ESGAR 2018 Book of Abstracts

ESGAR 2018 Book of Abstracts Insights into Imaging Education and strategies in European radiology ESGAR 2018 Book of Abstracts / Volume 9 / Supplement 2 / June 2018 ESGAR 2018 / June 12 – 15 / Dublin, Ireland th 29 Annual Meeting and Postgraduate Course t th h 2 29 9 A AN NN NU UA AL M L ME EE ET TIIN NG G A AN ND P D PO OS ST TG GR RA AD DU UA AT TE C E CO OU UR RS SE E BOOK OF ABSTRACTS INCLUDES ABSTRACTS OF SCIENTIFIC PRESENTATIONS J J J J J J J JU U U U U U U UN N N N N N N NE 1 E 1 E 1 E 1 E 1 E 1 E 1 E 12 2 2 2 2 2 2 2 - - - - - - - - 1 1 1 1 1 1 1 15 5 5 5 5 5 5 5 Eur European opean E ES SG GA AR R 2 20 01 18 8 Society Society of Gastr of Gastrointestinal ointestinal D D D D D D D DU U U U U U U UB B B B B B B BL L L L L L L LI I I I I I I IN N N N N N N N and Abdominal and Abdominal IIIIIIIIIIR R R R R R R R R RE E E E E E E E E EL L L L L L L L L LA A A A A A A A A AN N N N N N N N N ND D D D D D D D D D Radiology Radiology IMPORTANT ADDRESSES / CME / CONTRIBUTING SOCIETIES / CORPORATE MEMBERS ORGANISING SECRETARIAT CONFERENCE VENUE Central ESGAR Office The Convention Centre Dublin Esslinggasse 2/3 Spencer Dock AT – 1010 Vienna North Wall Quay Phone: +43 1 535 89 27 IE - D01 T1W6 Dublin 1 Fax: +43 1 535 89 27 -15 E-Mail: office@esgar.org WEBSITE www.esgar.org CME CREDITS th The “ESGAR 2018 29 Annual Meeting and Postgraduate The “ESGAR European Society of Gastrointestinal and Course, Dublin, Ireland, 12/06/2018-15/06/2018” has th Abdominal Radiology” (or) “ESGAR 2018 – 29 ANNUAL been accredited by the European Accreditation Council for MEETING AND POSTGRADUATE COURSE” is accredited Continuing Medical Education (EACCME®) with 25 by the European Accreditation Council for Continuing European CME credits (ECMEC®s). Each medical Medical Education (EACCME) to provide the following specialist should claim only those hours of credit that he/ CME activity for medical specialists. The EACCME is an she actually spent in the educational activity. institution of the European Union of Medical Specialists (UEMS), www.uems.net. CONTRIBUTING SOCIETIES SPONSORS ESGAR wishes to gratefully acknowledge the support of its Corporate Members: The Final Programme of ESGAR 2018 is available on the ESGAR Website www.esgar.org Date of publishing: June 2018 S653 $/)ªS  8-0634-1 TABLE OF CONTENTS TABLE OF CONTENTS Scientific Sessions, Wednesday, June 13 (SS 1 – SS 5) S658-S671 Scientific Sessions, Thursday, June 14 (SS 6 – SS 10) S672-S685 Scientific Sessions, Friday, June 15 (SS 11 – SS 15) S686-S698 Authors’ Index S699-S704 S654 654 COMMITTEES ESGAR 2018 MEETING PRESIDENT ESGAR EXECUTIVE COMMITTEE PRESIDENT Prof. Helen Fenlon S. Halligan, London/UK University College Dublin Department of Radiology PRESIDENT-ELECT Mater Misericordiae University Hospital and BreastCheck R.G.H. Beets-Tan, Amsterdam/NL 36 Eccles Street IE – Dublin 7 VICE PRESIDENT A. Laghi, Rome/IT ESGAR 2018 PROGRAMME COMMITTEE SECRETARY/ TREASURER M. Zins, Paris/FR CHAIRMEN M. Zins, Paris/FR PAST PRESIDENT C. Matos, Lisbon/PT C. Matos, Lisbon/PT MEMBERS EDUCATION COMMITTEE M.A. Bali, London/UK S. Jackson, Plymouth/UK A. Ba-Ssalamah, Vienna/AT MEMBERSHIP COMMITTEE R.G.H. Beets-Tan, Amsterdam/NL M.A. Bali, London/UK G. Brancatelli, Palermo/IT H. Fenlon, Dublin/IE RESEARCH COMMITTEE S. Halligan, London/UK J. Stoker, Amsterdam/NL S. Jackson, Plymouth/UK A. Laghi, Rome/IT WORKSHOP COMMITTEE G. Brancatelli, Palermo/IT J. Stoker, Amsterdam/NL V. Vilgrain, Clichy/FR MEETING PRESIDENT 2018 C.J. Zech, Basel/CH H. Fenlon, Dublin/IE PRE-MEETING PRESIDENT 2019 ESGAR 2018 LOCAL ORGANISING COMMITTEE A. Laghi, Rome/IT R. Browne, Dublin/IE MEMBERS AT LARGE C. Cronin, Dublin/IE A. Ba-Ssalamah, Vienna/AT T. Geoghegan, Dublin/IE V. Vilgrain, Clichy/FR M. Keogan, Dublin/IE C.J. Zech, Basel/CH M. Lee, Dublin/IE M. Maher, Cork/IE ESGAR EXECUTIVE DIRECTOR D.E. Malone, Dublin/IE B. Lindlbauer, Vienna/AT M. McNicholas, Dublin/IE M. Morrin, Dublin/IE LANGUAGE ABSTRACT EDITOR J. Murphy, Galway/IE C. Clarke, Nottingham/UK A. Ryan, Waterford/IE M. Ryan, Cork/IE T. Scanlon, Limerick/IE N. Sheehy, Dublin/IE M. Shelly, Limerick/IE S.J. Skehan, Dublin/IE M. Staunton, Cork/IE S655 ABSTRACT REVIEWING PANEL O. Akhan, Ankara/TR P. Prassopoulos, Alexandroupolis/GR C. Aubé, Angers/FR E. Quaia, Edinburgh/UK M.A. Bali, London/UK G.A. Rollandi, Genova/IT I. Bargellini, Pisa/IT M. Ronot, Clichy/FR T.V. Bartolotta, Palermo/IT W. Schima, Vienna/AT A. Ba-Ssalamah, Vienna/AT S. Schmidt Kobbe, Lausanne/CH R.G.H. Beets-Tan, Amsterdam/NL A. Schreyer, Regensburg/DE E. Biscaldi, Genova/IT O. Seror, Bondy/FR A. Blachar, Tel Aviv/IL S.J. Skehan, Dublin/IE G. Brancatelli, Palermo/IT M. Staunton, Cork/IE D.J. Breen, Southampton/UK S. Stojanovic, Novi Sad/RS F. Caseiro Alves, Coimbra/PT J. Stoker, Amsterdam/NL N. Courcoutsakis, Alexandroupolis/GR J.P. Tasu, Poitiers/FR L. Crocetti, Pisa/IT S.A. Taylor, London/UK L. Curvo-Semedo, Coimbra/PT S. Terraz, Geneva/CH R. Dondelinger, Liège/BE D.J.M. Tolan, Leeds/UK M. D’Onofrio, Verona/IT C. Triantopoulou, Athens/GR H. Fenlon, Dublin/IE V. Valek, Brno/CZ A. Furlan, Pittsburgh, PA/US V. Vilgrain, Clichy/FR Y. Gandon, Rennes/FR M.-P. Vullierme, Clichy/FR V. Goh, London/UK D. Weishaupt, Zurich/CH S. Gourtsoyianni, London/UK G.A. Zamboni, Verona/IT S. Gryspeerdt, Roeselare/BE C.J. Zech, Basel/CH L. Guimaraes, North York, ON/CA M. Zins, Paris/FR J.A. Guthrie, Leeds/UK S. Halligan, London/UK A. Hatzidakis, Heraklion/GR T. Helmberger, Munich/DE F. Iafrate, Rome/IT M. Karcaaltincaba, Ankara/TR N. Kartalis, Stockholm/SE H.-U. Laasch, Manchester/UK A. Laghi, Rome/IT M. Laniado, Dresden/DE J.M. Lee, Seoul/KR M. Lewin, Paris/FR O. Lucidarme, Paris/FR M. Maas, Amsterdam/NL A. Madureira, Porto/PT M. Maher, Cork/IE D.E. Malone, Dublin/IE T. Mang, Vienna/AT V. Maniatis, Aabenraa/DK D. Marin, Durham, NC/US L. Martí-Bonmatí, Valencia/ES C. Matos, Lisbon/PT Y. Menu, Paris/FR G. Morana, Treviso/IT G.H. Mostbeck, Vienna/AT A. Palkó, Szeged/HU P. Paolantonio, Rome/IT N. Papanikolaou, Lisbon/PT R. Pozzi Mucelli, Verona/IT S656 Scientific Sessions Authors’ Index S657 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 1.3 11:00 - 12:30 The Auditorium Radiological-pathological correlation of pre-transplant imaging and explanted livers: what is the prevalence of Scientific Session SS 1 HCC in patients with liver imaging reporting and data Focal liver lesions: new trends in imaging system 2017 category 5, 4/5 and 4 lesions? of HCC A.C. O‘Brien, R. Keane, B. Pierce, C. O‘Brien, N. Nolan, P.A. McCormick, D.E. Malone, S.J. Skehan, R. Ryan, R. Gibney; Dublin/IE SS 1.1 Comparison of European association for the study of the Purpose: The Irish National Centre for Liver Transplantation (OLT) serves a pop- liver, liver imaging and reporting data system criteria and ulation of 4.5 million. Liver imaging reporting and data system (LI-RADS) catego- radiological opinion for diagnosing HCC in cirrhotic livers ry designation is made at a multidisciplinary meeting by 1 or more of 4 abdominal 1 2 2 2 C.G.D. Clarke , R. Albazaz , C.R. Smith , I. Rowe , radiologists. The aim of this study is to evaluate the prevalence of HCC in OLT 2 2 2 2 D. Treanor , J.I. Wyatt , M.B. Sheridan , J.A. Guthrie ; patients whose pre-OLT imaging showed LI-RADS 4 or 5 lesion(s). 1 2 Nottingham/UK, Leeds/UK Material and methods: OLT patients with a pre-operative diagnosis of HCC and explants with HCC found at pathology from 01/01/2014 to 09/09/2017 were Purpose: Imaging criteria for the non-invasive diagnosis of HCC have been es- identified using the OLT database. Corresponding pre-OLT imaging reports (CT/ tablished by many organisations. The reported diagnostic performance of MRI in MRI) were reviewed. In patients with LI-RADS 2017 class 5, 4/5 or 4 lesions, the the diagnosis of HCC is widely variable. Our aim was to determine the sensitivity histopathology coding of explanted livers was reviewed. Data were collated in an and specificity of these criteria by correlating with a histological diagnosis from MSExcel spreadsheet and analysed on a per patient basis in 2 groups (LI-RADS whole liver explants. 5 and LI-RADS 4/5 or 4). Material and methods: This is a single-centre, retrospective review. Participants Results: 53 OLT patients met selection criteria. Pre-operative imaging reported were selected based on the following: consecutive adults (>=18yrs) listed for LI-RADS 5, 4/5 or 4 lesions in 41 of 53 explanted livers. LI-RADS 5: n = 36 pa- liver transplantation in 2014/2015, with cirrhosis at the time of MR scanning with tients (7 had pre-OLT TACE), 35 had HCC, 1 had macroregenerative cirrhosis hepatocyte-specific contrast agent, and at least one liver lesion >=1cm on MR without HCC; 1 also had a collision tumour (HCC/cholangiocarcinoma). HCC with histology from subsequent liver explant for comparison. Patients were ran- prevalence: 97%. LI-RADS 4/5: n =1 patient, HCC confirmed at OLT. LI-RADS 4: domly allocated to two independent consultant radiologists for review. Each le- n = 4 patients, 3 had HCC and dysplastic nodules only were identified in the sion was assessed against international criteria [European association for the other patient. HCC prevalence 80%. study of the liver (EASL), liver imaging and reporting data system (LIRADS)], and Conclusion: The prevalence of HCC in OLT patients with LI-RADS 2017 class 5 given a ‚radiologist impression‘ score of 1-5 (1=definitely benign, 5=definitely lesions is acceptably high. The small LI-RADS 4/5 or 4 group had a lower HCC HCC). prevalence. More data are needed for optimal management of these patients. Results: Total of 268 patient records were reviewed, with 118 eligible lesions identified from 50 patients. Median lesion size was 15.5 (interquartile range 12- 21) mm. Mean age 56 (+/-7.7) yrs with M:F ratio of 4:1. Sensitivity, specificity and SS 1.4 PPV for EASL was 0.47, 0.82 and 0.83, for LIRADS LR5 was 0.39, 0.88 and 0.85, Prospective intraindividual comparison of gadoxetic acid for LIRADS LR4+5+TIV was 0.73, 0.69, and 0.81, and for ‚radiologist impression‘ and gadoterate meglumine for the diagnosis of HCCs with of probably or definitely HCC, available in 108 lesions, was 0.79, 0.79 and 0.88. liver imaging reporting and data system 1 2 1 1 2 Conclusion: MR imaging has moderate sensitivity and good specificity in the Y.K. Kim , J.H. Min , W.K. Jeong ; Seoul/KR, Daejeon/KR diagnosis of HCC and there is considerable variation depending on the criteria Purpose: This prospective study intraindividually compared the efficacy of ga - used. ‘Radiologist impression’ has better sensitivity when compared to LIRADS doxetic acid (Gd-EOB-DTPA)-enhanced MRI and gadoterate meglumine (Gd- and EASL. DOTA)-enhanced MRI for the diagnosis of HCC with the liver imaging reporting and data system (LI-RADS). SS 1.2 Material and methods: Between November 2016 and November 2017, we pro- spectively included 91 patients with chronic liver disease who underwent both withdrawn by the authors Gd-EOB-DTPA-MRI and Gd-DOTA-MRI (interval range, 9 to 30 days) for first detected hepatic nodule on US: 107 lesions (95 HCCs, 2 cholangiocarcinomas, 7 hemangiomas, 2 dysplastic nodules, 1 neuroendocrine tumor; size range: 0.8- 3.0 cm) were identified by surgical resection. Two observers reviewed two MRIs based on the LI-RADS v2017. Results: We found a tendency toward higher sensitivity (observer 1, 76.8% vs 57.9% and observer 2, 81.1% vs. 61.3%) and accuracy (observer 1, 81.2% vs 65.8% and observer 2, 84.6% vs. 68.7%) with LR-5 category on Gd-DOTA-MRI than with that on Gd-EOB-DTPA-MRI (P < 0.001). When applying arterial hyper- enhancement with either washout on 3-min Gd-EOB-DTPA MRI or isointensity with capsule on 3 min Gd-DOTA-MRI, 16 HCCs on Gd-EOB-DPTA-MRI and 10 HCCs on Gd-DOTA-MRI were upgraded to LR-5 category while achieving 100% specificity with both MRIs. One cholangiocarcinoma was correctly classified with Gd-EOB-DTPA MRI due to targetoid appearance. Conclusion: In the diagnosis of HCC with LI-RADS, MRI using extracellular con- trast agent showed better sensitivity and accuracy than Gd-EOB-DTPA-MRI al- though they showed 100% specificity. We could achieve better diagnostic per - formance with applying washout on 3 min Gd-EOB-DTPA-MRI or isointensity with capsule on 3-min Gd-DOTA MRI than conventional criteria, while maintain- ing 100% specificity. S658 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 1.5 SS 1.7 Prospective evaluation of dynamic MR with gadoxetic Hypervascular tumors in patients with primary Budd- acid for the non-invasive diagnosis of HCC in newly Chiari syndrome: is the washout sign specific for HCC? detected nodules An MRI study C. Ayuso, A. Darnell, J. Rimola, A. García-Criado, R. Vilana, A. M. Van Wettere, O. Bruno, P.-E. Rautou, A. Plessier, Forner, J. Bruix; Barcelona/ES V. Vilgrain, M. Ronot; Clichy/FR Purpose: To evaluate the diagnostic accuracy of gadoxetic acid-MR (EOB-MR) Purpose: To reappraise the imaging characteristics of focal liver lesions (FLL) in for non-invasive HCC diagnosis in liver nodules <2 cm detected by screening patients with Budd-Chiari syndrome (BCS), with a focus on the diagnostic value US. of washout for the differentiation between benign and malignant tumors. Material and methods: Cirrhotic patients with newly detected solitary nodules Material and methods: Between 2000 and 2016, patients with a diagnosis of <2 cm by US were included. After written informed consent, extracellular con- BCS and FLL on MRI were included. MRI was retrospectively reviewed by two trast-MR (EC-MR) and EOB-MR were performed in less than 1 month. Final di- radiologists blinded to the nature of these lesions. Patients and tumor character- agnosis was based on the validated European Association for the Study of the istics were recorded, with a focus on signal intensity on MRI sequences. HCC Liver (EASL)/American Association for the Study of Liver Diseases (AASLD) and benign tumors were compared using Fisher‘s test or a Chi-squared test and guidelines criteria (specific vascular profile on EC-MR or biopsy). Blind, double the Student‘s t test or Mann-Whitney test. reading was performed in EOB-MR studies. Criteria for HCC diagnosis using Results: 49 patients (34 female, 69%) with 241 benign lesions and 12 HCC were EOB-MR were arterial contrast uptake and either washout in the portal phase or analyzed. Patients with HCC were significantly older (mean 44±16 vs. 33±9, p hypointensity in the hepatobiliary phase (HBp). =.005), with higher baseline serum alpha-fetoprotein (AFP) level (median 16 [2- Results: 62 consecutive cirrhotic patients Child-Pugh A (53) or B (9) were in- 9000] vs. 3 [2-25] ng/mL, p =.007). HCC was significantly larger (mean 32±16 vs. cluded. Final diagnosis was: 41 (66.1%) HCC, 2 intrahepatic cholangiocarcino- 11±5mm, p<0.001), more frequently hypointense on T1-weighted (58% vs. 4%, ma (ICC), 1 colo-rectal cancer metastasis, and 18 benign conditions. Patients p=.001) and hyperintense on T2-weighted images (58% vs. 18%, p=.003) than with benign lesions were followed during a median of 23 months to discard ma- benign lesions. 28% of the benign lesions showed both hypervascularization and lignancy. EC-MR diagnosed 26 out of 41 HCC nodules (sensitivity 63.4%, washout too. In lesions >10mm, the sensitivity/specificity of these features was CI95%: 46.9-77.9). The sensitivity and specificity for EOB-MR were 56.1% 75%/67%. Adding signal hyperintensity on T1-w images raised the specificity to (CI95%: 39.7-71.5) and 90.5% (CI95%: 69.8-98.8), respectively, with a positive 82%. predictive value (PPV) of 92 (CI95%: 74-99) and negative predictive value (NPV) Conclusion: Washout appearance was observed in close to one-third of benign of 51.4 (CI95%: 34.4-68.1). False positives were observed in the two ICC. In lesions leading to an unacceptably low specificity of this feature for the diagnosis EOB-MR, wash-in was present in 82.9% and portal wash-out in 42%. Low signal of HCC. Other imaging ancillary features and patients characteristics should be intensity in the 20-minute HBp was observed in 63.4% of HCC nodules, whereas carefully looked at, especially pre-contrast signal intensity and serum AFP level. the remaining 15 HCCs were iso- or hyperintense. Conclusion: EOB-MR is not superior to EC-MR for non-invasive diagnosis of SS 1.8 HCC in nodules <2 cm in cirrhotic patients. Liver imaging reporting and data system v2017 categorization of HCC using multiphasic MDCT: SS 1.6 comparison between patients with and without moderate Proliferation status of HCC predicted by whole-lesion to severe fatty liver 1 1 1 2 texture analysis on gadolinium ethoxybenzyl S.S. Kim , J.A. Hwang , H.C. Shin , S.-Y. Choi ; 1 2 diethylenetriamine pentaacetic acid-enhanced MRI Cheonan-si, Chungcheongnam-do/KR, Bucheon/KR Z. Ye; Chengdu/CN Purpose: To compare the sensitivity of the liver imaging reporting and data sys- Purpose: To explore the potential association between whole-lesion texture fea- tem (LI-RADS) v2017 for categorizing HCC using multiphasic MDCT between tures on gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB- patients with and without moderate to severe fatty liver (MSFL). DTPA)-enhanced MRI and the Ki67 labeling index (Ki67 LI) in HCC patients. Material and methods: This retrospective study was approved by the institu- Material and methods: 61 consecutive surgically confirmed HCC patients were tional review board, and the informed consent requirement was waived. A total prospectively examined on a 3.0 T scanner. Texture features on individual se- of 106 high-risk patients with 112 pathologically proven HCCs who underwent quence, including pre-contrast T1-weighted (T1-pre), T2-weighted (T2WI), arte- multiphasic MDCT were included. Patients were classified into MSFL (n=26) and rial phase (AP), portal venous phase (PVP) and hepatobiliary phase (HBP) imag- non-MSFL (n=80) group according to unenhanced CT liver and spleen parenchy- ing were derived using an in-house software (Analysis Kit, GE Healthcare). The mal attenuation. Two independent radiologists assigned LI-RADS categories and Ki67 LI was histopathologically determined and classified into low-proliferation accessed HCC features on MDCT. Sensitivities of LR-5/tumor in vein (TIV) and (Ki-67 LI<15%) and high-proliferation group (Ki-67 LI>15%). Lasso model was frequencies of major HCC features on MDCT were compared between two adopted for feature selection. Differences in texture features between low- and groups. high-proliferation groups were compared. Diagnostic performance of each se- Results: Sensitivities of LR-5/TIV were not significantly different between MSFL lected texture parameter in differentiating high-proliferation from low-prolifera- and non-MSFL group (73.1% [19/26] vs. 83.7% [72/86], P = 0.255; 80.8% [21/26] tion group was evaluated by receiver operation characteristic (ROC) analysis. vs. 87.2% [75/86], 0.522 in reviewers 1 and 2, respectively). No significant differ - Results: The cluster prominence and high grey level run emphasis of T1-pre, the ences in frequencies of arterial hyperenhancement, washout, and capsule were Haralick correlation and inverse difference moment (IDM) of T2WI, the IDM of AP, observed between the two groups (96.2% [25/26] vs. 98.8% [85/86], P = 0.412; the cluster shade and high grey level run emphasis of PVP and the correlation of 80.8% [21/26] vs. 89.5% [77/86], P = 0.308; and 53.8% [14/26] vs. 57% [49/86], HBP showed significant differences between low- and high-proliferation groups P = 0.778, respectively). Two HCCs (7.7% [2/26]) in MSFL group showed higher (P<0.05). The largest areas under ROC curve (AUCs) of T1-pre (high grey level attenuation compared with background liver on unenhanced image. run emphasis), T2WI (Haralick correlation), AP (IDM), PVP (high grey level run Conclusion: LI-RADS using MDCT showed comparable sensitivity for the diag- emphasis) and HBP (correlation) were 0.76 (CI[0.63, 0.86]), 0.70 (CI[0.57, 0.81]), nosis of HCCs regardless of MSFL. 0.69 (CI[0.56, 0.87]), 0.71 (CI[0.58, 0.82]) and 0.70 (CI[0.57, 0.81]), respectively, for identifying high-proliferation HCC. Correlation of HBP showed highest sensi- tivity (91.43%, CI[76.9%, 98.2%]) while IDM of AP showed highest specificity (96.15%, CI[80.4%, 99.9%]). Conclusion: Whole-lesion texture features on Gd-EOB-DTPA-enhanced MRI are promising in predicting and characterizing proliferation status of HCC. S659 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 The Liffey A SS 1.9 Pre-treatment estimation of liver function using T1 Scientific Session SS 2 mapping of gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid-enhanced MRI and normal liver volume Insights in imaging of rectal cancer in patients with HCC T. Duan, L. Cao, M. Wang, B. Song; Chengdu/CN SS 2.1 Purpose: This study aimed to prospectively evaluate liver function in patients The standard of MRI rectal cancer staging reporting in with HCC using T1 mapping before and after 20 minutes of gadolinium-eth- clinical practice: a case for standardization? oxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) injection and P.J. Brown, H. Rossington, D.J.M. Tolan; Leeds/UK normal liver volume, compared with indocyanine green (ICG) retention at 15 Purpose: Rectal cancer staging with magnetic resonance imaging (RCS-MRI) min (ICG R-15). allows accurate assessment of tumours. Significant variability exists in the Material and methods: 101 consecutive patients with HCC were included in content of reports. Template-style reporting can improve reporting standards, this study. T1 relaxation time of the livers before and 20 minutes after Gd-EOB- but its use is not widespread. Given the implications for treatment we have DTPA injection (pre-T1 and post-T1) was measured, and the reduction (Δ) and evaluated standard practice amongst radiologists to measure the current qual- reduction rate of T1 relaxation time (Δ%) were calculated. Volumes of normal ity of RCS-MRI reports. liver parenchyma were conducted and the products of T1 parameters and vol- Material and methods: 16 UK colorectal cancer multidisciplinary teams ume were calculated. ICG R-15 was also recorded. Liver function insufficiency (MDTs) serving a population over 5 million in Yorkshire were invited to submit a was defined as ICG R-15 >10. T1 parameters, volumes and the products were maximum of 10 consecutive RCS-MRI reports from January 2016 for each correlated with both ICG R-15 using Spearman rank correlation analysis. Re- radiologist participating in the colorectal MDT. All radiologists were subspe- ceiver operation characteristic (ROC) analysis of predicting the liver function cialists in GI imaging with training in RCS-MRI. Reports were compared to a insufficiency was performed for T1 parameters, volumes and the products. reference standard based on key data points in The Union for International Results: The post-T1 (r=0.450, P<0.001), Δ (r=-0.287, P=0.011), Δ% (r=-0.419, Cancer Control (UICC)-TNM 5 staging and other recognised factors influenc - P<0.001) volume (r=-0.374, P<0.001), product of Δ and volume (r=-0.442, ing case management. P<0.001) as well as product of Δ% and volume showed weak to moderate Results: 293 RCS-MRI reports were submitted from 32 of 42 (76%) eligible correlations with ICG R-15. In predicting the liver function insufficiency, the radiologists. 81 of 293 (28%) reports used a template. Template report usage area under the ROC of post-T1, Δ, Δ%, volume, product of Δ and volume and significantly increased recording of key data points versus non-template re - product of Δ% and volume were 0.767, 0.619, 0.722, 0.745, 0.791 and 0.856, ports for extra-mural vascular invasion (EMVI) status (98.8% v 53.3%, p < respectively. 0.001) and circumferential-resection margin (CRM) status (98.8% v 63.2%, p < Conclusion: A combination of T1 mapping and normal liver volume can help 0.001). Local tumour stage (97.5% v 92.0%, NS) and nodal status (98.8% v in evaluating liver function. 95.8%, NS) were usually reported with similar frequency. Conclusion: Non-template reporting in RCS-MRI excludes critical variables SS 1.10 that influence clinical decisions for rectal cancer. Standardised template re - Combining CT-liver perfusion and MRI with hepato- porting significantly improves collection of key data. specific contrast agent to increase diagnostic accuracy in patients with suspected HCC: work in progress SS 2.2 A.A. Hatzidakis, G. Kalarakis, K. Perisinakis, E. Chryssou, Multi-exponential T2* mapping distinguishes benign A. Papadakis, A. Karantanas; Heraklion/GR from malignant lymph nodes in rectal cancer patients: Purpose: Gadoxetic acid-enhanced MRI (GaE-MRI) is used for HCC surveil- an ex vivo and in vivo experiment 1 1 2 1 lance. We sought to compare the diagnostic accuracy of CT-liver perfusion I. Santiago , J. Santinha , A. Ianus , N. Papanikolaou , 1 1 1 2 (CTLP) plus GaE-MRI versus GaE-MRI alone for detection and characteriza- C. Matos , N. Shemesh ; Lisbon/PT, London/UK tion of suspected HCC lesions. Purpose: To investigate multi-exponential decay in multi-gradient-echo (MGE) Material and methods: Sixteen patients (all male, 12 cirrhotic) under HCC MRI for benign/malignant lymph node (LN) distinction in rectal cancer patients, surveillance before or after percutaneous HCC treatment, underwent GaE-MRI both ex vivo at 16.4T and in vivo at 1.5T. and CTLP (with a maximum interval of 6 weeks). In total, 16 pairs of CTLP and Material and methods: Ex vivo experiment: 33 benign and 32 malignant LNs GaE-MRI examinations were studied. GaE-MRI, was performed on a 1.5T- were retrieved from 11N+ rectal cancer patient specimens, preserved in 4% system (Siemens Vision-Hybrid). HCCs were characterized according to con- formaldehyde, moved to 1% phosphate buffered saline (PBS) for 24h, im- ventional MRI criteria. CTLP-maps of mean slope of increase were generated mersed in Flourinert and acquired in a 16.4T Bruker scanner. A fat-suppressed using a 128-CT-system (GE Revolution HD) and the lesions were depicted MGE acquisition was acquired: 50 TEs (echo time) starting at 1.6ms; 1.4ms based on a previously determined cut-off value. Diagnoses based on GaE-MRI interval; repetition time (TR)=1500ms; flip angle (FA)=50º; slice thickness alone and GaE-MRI plus CTLP were compared with digital subtraction angiog- 2 (ST)=0.3mm; in-plane resolution=0.1x0.1mm ; bandwidth=125000Hz; 25 aver- raphy (DSA). ages. In vivo experiment: 8 rectal cancer patients underwent MGE during stag- Results: Of the total 44 lesions identified (median diameter 20mm, range ing on a 1.5T Philips scanner: 32 TEs starting at 2.37ms; 2.37ms interval; 5-124mm), 19 were characterized as HCCs by DSA. GaE-MRI identified 16 2 TR=1519ms; FA=55º; ST=4mm; in-plane resolution=0.42x0.42mm ; true-positive, 19 true-negative lesions and misdiagnosed 6 false-positive and bandwidth=431.3Hz; 2 averages. Six patients underwent total mesorectal ex- 3 false-negative lesions, providing 84.2% sensitivity and 76% specificity. The cision without neoadjuvant therapy and of the LNs retrieved, 36 benign and 27 combination of GaE-MRI and CTLP identified 19 true-positive, 24 true-nega - malignant were matched to MGE images. Data from ex vivo whole-node vol- tive and 1 false-positive lesion increasing sensitivity and specificity to 100% ume of interest (VOI) and in vivo single-slice whole-node region of interest and 96%, respectively (p=0.0025). Another 12-mm nodule, characterized as (ROI) were used for analysis. Datasets were fitted using 1-, 2- and 3-compart - HCC by CTLP/GaE-MRI, was initially not detected by DSA, but was revealed ment T2* models. Models were compared based on Bayesian information cri- in a subsequent one, 6 months later. terion. Histogram analysis was performed for ex vivo and receiver operating Conclusion: Combination of GaE-MRI and CTLP may increase small (<15mm curves (ROC) were computed. Mann-Whitney U test was employed for param- diameter) hepatic nodule characterization accuracy, enabling more efficient eter comparison. patient selection for early and individualized loco-regional treatment. Results: 2-compartment model ranked first in both datasets. For ex vivo, sig - nificant differences were found in metrics derived from T2*a, T2*b, f and ΔΩ, the most discriminative being P75 of f(p=0.002). For in vivo, T2*a showed the highest specificity for malignancy (0.94) and T2*b showed the highest discrim - inative power (auROC=0.76). The specificity of T2*a, T2*b and their combina - tion exceeded that of reported conventional imaging criteria. Conclusion: Our results indicate multi-compartment T2* mapping may be of added value for LN staging in rectal cancer. S660 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 2.3 SS 2.5 Comparison of 8 diffusion MR models to distinguish The tram track sign: a new, highly specific and reliable benign/malignant mesorectal lymph nodes from rectal sign for the detection of complete response after cancer patients performed ex vivo at ultrahigh field neoadjuvant therapy in rectal cancer 1 2 1 2 A. Ianus , I. Santiago , D. Alexander , C. Matos , I. Santiago, M.J. Barata, N. Figueiredo, O. Parés, C. Matos; 2 1 2 N. Shemesh ; London/UK, Lisbon/PT Lisbon/PT Purpose: To assess how different compartment models capture the signal be- Purpose: To assess the diagnostic value of the tram track sign to predict path- haviour and differentiate benign from malignant lymph nodes (LNs) retrieved ologic/sustained clinical complete response (CR) after neoadjuvant therapy in from rectal cancer patients‘ specimens, using a rich diffusion MRI protocol, at rectal cancer. 16.4T. Material and methods: We retreived from our prospectively organized rectal Material and methods: Twenty-six benign and 32 malignant LNs originating cancer database all patients who underwent neoadjuvant therapy followed by from 11 consecutive N+ patients who underwent surgery without neoadjuvant restaging MRI (+)10 weeks post-radiotherapy, between October 2013 and therapy were preserved in 4% formaldehyde, moved to 1% phosphate-buff- March 2017. Two dedicated radiologists blindly and independently reviewed ered saline (PBS) 24h before acquisition, immersed in Flourinert and imaged in T2 and diffusion-weighted images and graded response according to tumor a 16.4T Bruker scanner. Stimulated echo acquisition mode diffusion-weighted regression grade (mrTRG) and diffusion weighted imaging (mrDWI) using an imaging was employed using 4 different b values, varying diffusion gradient in-house ordinal scale from 1-5. They also assessed the presence/absence of times and 6 different gradient directions for each parameter combination. Data the tram track sign (mrTTS) - a double, uninterrupted, markedly T2-hypoin- were normalized for each diffusion time and fitted to 8 different diffusion mod - tense line at previous tumour location. Endoscopic data at same timepoint, els. Averaged signal over whole-node and up to 10 individual LN regions of graded according to an ordinal scale from 0-4, was also retrieved. For statisti- interest (ROIs), as delineated by a dedicated radiologist, were used. Models cal analysis, primary endpoints were to compare the accuracy of endoscopy, were compared based on goodness of fit. The benign vs malignant differentia - mrTRG, mrDWI and mrTTS for the identification of sustained complete re - tion ability of each model was analysed using general linear model (GLM) with sponders (SCR), defined as patients with pathologic CR or clinical CR at 1 binomial distribution and receiver operating characteristic (ROC) curves were year. computed. Results: 47 patients considered eligible: 26 males, mean age 63.8y. 20 under- Results: ZeppelinSphere was the model that ranked first in most instances, went surgery. No significant differences were found for endoscopy. For ob - whereas the Ball model (equivalent to ADC) ranked last in most instances. For server 1, significant differences were found between mrTRG /mrTRG 1-2 3- whole-node analysis, Tensor and ZeppelinBall models best differentiated be- (p=0.03) and mrTTS mrTTS (p=0.02), and for observer 2, between mrDWI / 5 +/ - 1-2 nign from malignant LNs (area under ROC=0.78 and 0.77, respectively). For mrDWI (p=0.03) and mrTTS mrTTS (p<0.01). Inter-rater agreement was poor 3-5 +/ - individual ROIs, models with restriction (BallSphereSphere and Zeppelin- for mrTRG[0.15(p=0.14)], moderate for mrDWI[0.55 (p<0.01)] and very good for Sphere) yielded better results. mrTTS[0.83(p<0.01)]. mrTTS provided a sensitivity of 0.56/0.61 for observers Conclusion: Accounting for restricted diffusion improved the data fit of diffu - 1/2, respectively, and a specificity/PPV of 1 for both readers. sion MRI signal in LNs from rectal cancer specimens, which may provide more Conclusion: mrTTS has very high specificity, positive predictive value and in - specificity towards tissue microstructure. It also impacted lymph node differ - terrater agreement for the identification of SCR in rectal cancer. Its sensitivity entiation. is, however, low; therefore, it should be combined with other post-neoadjuvant therapy assessment tools. SS 2.4 T2-weighted signal intensity to predict complete and SS 2.6 good response after neoadjuvant chemoradiation Patterns of fibrosis on rectal MRI in clinical complete therapy in patients with rectal cancer responders undergoing wait-and-see after 1 2 1 R.A.P. Dijkhoff , S.G. Drago , J. Van Griethuysen , chemoradiotherapy for rectal cancer: correlation with 1 3 1 D.M.J. Lambregts , F.C.H. Bakers , R.G.H. Beets-Tan , functional outcome 1 1 2 3 1 1 2 M. Maas ; Amsterdam/NL, Monza/IT, Maastricht/NL T.N. Boellaard , D.M.J. Lambregts , A. Delli Pizzi , 1 3 1 M. Van Der Sande , B.J.P. Hupkens , J. Van Griethuysen , Purpose: To determine whether T2-weighted signal intensity (T2W-SI) before 1 1 1 1 G.L. Beets , R.G.H. Beets-Tan , M. Maas ; Amsterdam/NL, and after neoadjuvant chemoradiation therapy (CRT) can predict complete and 2 3 Chieti/IT, Maastricht/NL good response in patients with rectal cancer. Material and methods: 171 consecutive patients with rectal cancer treated Purpose: To study whether morphologic patterns of fibrosis on T2W-MRI in with neoadjuvant CRT underwent MRI before and 8-10 weeks post-CRT. The non-operated clinical complete responders after chemoradiotherapy (CRT) for primary tumour and remnant after CRT were manually delineated on T2W-MRI. rectal cancer are related to long-term functional outcome. Histogram analyses were performed on these volumes of interest with pyradi- Material and methods: N=68 patients were retrospectively analysed. All had omics. Extracted parameters were mean, median, standard deviation (SD), a sustained complete response and underwent MRI follow-up (FU) 3-6 month- range, minimum, maximum T2W-SI before and after CRT. Change between ly as part of wait-and-see program (median FU 34 months (range 16-62)). Mor- pre- and post-CRT T2W-SI parameters was calculated. Heterogeneity of T2W- phology of the rectal wall was assessed (2-reader consensus) on the T2W SI was assessed by the coefficient of variance (CoV=SD/mean). T2W-SI pa - FU-MRIs and classified according to 4 patterns: [1] no fibrosis (=normalised rameters were compared between complete (CR; ypT0), good responders rectal wall), [2] minimal fibrosis, [3] full-thickness fibrosis or [4] irregular/spicu - (GR; ypT0-1) and non-responders (NR; ypT2-4). Reference standard was ei- lated fibrosis. Long-term functional outcome was assessed with the Vaizey ther histology after surgery or a wait-and-see programme with at least 2 years score, a 0-24 point score on bowel function/incontinence, as the main out- of recurrence-free follow-up. come. Vaizey-scores were compared between the 4 patterns of fibrosis. Results: 36/135 patients had a CR and 49/171 a GR. Pre-CRT maximum SI Results: 5 patients showed no fibrosis, 45 minimal, 15 full-thickness and 3 ir - (941.62 vs. 1149.63, p=0.028), heterogeneity of the SI (0.265 vs. 0.291, regular/spiculated fibrosis. Mean Vaizey score was 1.6 for the patients with no p=0.017) and SI range (828.56 vs. 1074.52, p=0.009) were significantly lower fibrosis versus 4.5/3.9/4.0 for the patients with minimal/full-thickness/irregular in CR than in NR. Pre-CRT minimum SI was significantly higher in both CR fibrosis, respectively (P=0.557). For the patients (n=38) with a distal rectal tu - (113.06 vs. 75.10, p=0.017) and GR (117.02 vs. 69.46, p=0.001) compared to mour (≤3 cm from the anorectal junction), mean Vaizey score was 1.3 in the no NR. Mean difference in SI between primary and restaging was larger in both fibrosis group versus 4.7 for the minimal fibrosis and 6.5 for the full-thickness CR and GR compared to NR (CR: -14257 vs. NR: -105.44, p=0.114 and GR: fibrosis group (no patients had irregular fibrosis; P=0.198). 176.83 vs. NR: -87.72, p=0.001). Conclusion: Patients with a clinical complete response after CRT who show a Conclusion: The pre-CRT T2W-SI is significantly higher in CR and GR with a normalised rectal wall without fibrosis appear to have a better functional out - larger decrease in mean SI after CRT. These parameters could be a potential come in terms of incontinence and bowel function compared to patients with non-invasive marker for predicting complete/good response in patients with fibrosis. rectal cancer before and after CRT. S661 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 2.7 SS 2.8 18F-fluorodeoxyglucose positron emission tomography Ulcerated scars post neoadjuvant therapy in rectal ([18F]-FDG PET)/MRI in locally advanced rectal cancer cancer: can morphologic changes and diffusion- after preoperative chemo-radiotherapy: a comparison weighted imaging restriction patterns predict a with conventional imaging complete response? F. Crimì, C. Lacognata, I. Maretto, A. Perin, D. Cecchin, M.J. Barata, I. Santiago, N. Figueiredo, O. Parés, C. Matos; P. Zucchetta, F. Pomerri; Padua/IT Lisbon/PT Purpose: To assess the accuracy of restaging after preoperative chemoradio- Purpose: Are changes in angle, thickness and pattern of restriction to diffusion therapy (pCRT) with whole-body PET/MRI, pelvic T2-weighted (T2W) plus dif- of the tumour scar predictive of complete response in patients with rectal can- fusion-weighted imaging (DWI) MRI (pelvic-MRI), pelvic T2W MRI (standard- cer presenting with an ulcer at tumour site on endoscopy after neoadjuvant MRI) and thoracoabdominal CT when predicting histopathologic TNM stage in therapy (NAT)? locally advanced rectal cancer (LARC). Material and methods: All patients with rectal cancer who underwent NAT Material and methods: 26 patients with LARC underwent PET/MRI and CT and presented with an ulcerated scar on restaging endoscopy were retrieved before and after pCRT for TNM staging. 21 were treated with total mesorectal from our database. MRIs were reviewed independently and blindly by 2 radi- excision and 5 with transanal local excision. Histopathologic findings or a fol - ologists, which recorded, in 4 different timepoints post-radiotherapy: 1st low-up of at least 1-year were the reference standards. One radiologist evalu- (mean: 8,6 weeks), 2nd (mean: 20,7 weeks), 3rd (mean: 33,4 weeks) and last ated pelvic MRI and CT. A second radiologist evaluated standard MRI. A third (mean: 81,2 weeks): the angle between deepest point and inner borders of ul- radiologist and a nuclear medicine physician assessed PET/MRI. T staging cerated scar (ANG); the scar thickness (STh); and the distribution pattern of results were grouped in T0 and residual disease (T≥1). N stage was classified high-signal intensity on diffusion-weighted imgaing (DIST). Primary outcome on a per-patient basis as positive or negative using MRI dimensional criteria was presence/absence of signs of viable tumour at pathology or ≥1-year fol- (≥5mm per node), MRI lymph node global size reduction rate criteria (reduc- low-up (including regular digital rectal examination, endoscopy and MRI), pa- tion<70%) and PET/MRI dimensional criteria and/or nodal FDG uptake. tients considered non-complete responders (NCR) and complete responders Results: Sensitivity and specificity for ypT0 were 100%-85.7% for PET/MRI, (CR), respectively. Mann-Whitney-U and Fisher´s exact test were used for sta- 94.7%-85.7% for pelvic-MRI and 94.7%-57.1% for standard-MRI. For ypN+ tistical analysis. Interobserver agreement was assessed using intraclass cor- with dimensional criteria, sensitivity and specificity were 100%-88.9% for PET/ relation coefficient (ICC). MRI and 75%-88.89% for pelvic-MRI. The values for pelvic-MRI changed to Results: 15 patients considered eligible (mean age= 66y; 8 men), 7 underwent 87.5% and 72.2% using lymph node global size reduction rate criteria. PET/ surgery, 6 had persistent disease at pathology. Significant differences were MRI correctly diagnosed two liver and one distant nodal metastases while found for both readers in percentage change of ANG and STh (median values): missed a lung metastasis. ANG1st_ANG3rd: Reader1:NCR=+9,2%; CR=-24,7%, p=0.02; Reader2: Conclusion: PET/MRI improves the accuracy of ycTN staging compared to NCR=+44%; CR=-18,2%, p<0.01; ICC:0,44; ANG2nd_ANG3rd: Reader1: MRI, but performs worse than CT in ycM staging. Initial results are promising; NCR=+3,1%; CR=-16,2%, p=0.01; Reader2: NCR=+11,5%; CR=-13,8%, however, a larger cohort of patients should be examined introducing sequenc- p=0.04; ICC:0.34; ANG1st_ANGlast: Reader1: NCR=+15,8%; CR=-24%, es for lung and gadolinium for liver metastases. p=0.03; Reader2: NCR=+45,3%; CR=-30%, p=0.04; ICC:0.70; STh1st_ STh2nd: Reader1: NCR=0%;CR =-14,9%; p<0.01; Reader2: NCR=0%; CR=- 18,3%; p=0.04; ICC:0,80; STh2nd_STh3rd: Reader1: NCR=5,1%; CR=-9,8%; p<0.01; Reader2: NCR=+40%; CR=-6,3%; p=0.04; ICC:0,48; STh1st_STh3rd: Reader1: NCR=0%; CR=-29,6%; p=0.02; Reader2: NCR=22,2%; CR=-26,8%; p<0.01; ICC:0,96; DIST was significantly different for both readers at last time - point (absent/endoluminal linear vs intramural/irregular; p=0.01, ICC: 0.88). Conclusion: Patients with an ulcerated scar on endoscopy after NAT were more likely to be CRs when progressive contraction and thinning of the scar was observed on follow-up MRIs, and when an absent/endoluminal linear pat- tern of restriction to diffusion was observed at last follow-up. S662 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 Liffey Hall 2 SS 2.9 Endorectal ultrasound is accurate for the assessment of Scientific Session SS 3 anterior resection margin in low rectal cancer 1 2 1 1 1 J. Shur , D. Burling , A. Corr , J.T. Jenkins ; London/UK, Imaging of acute abdominal conditions Harrow/UK Purpose: Accurate assessment of anterior resection margin (ARM) in low rec- SS 3.1 tal cancer (LRC) helps ensure an R0 excision and appropriate triage to neo- Can volumes of extrapancreatic inflammation and adjuvant treatment (naCRT), and with MRI can be difficult due to paucity of pancreas on abdominal computed tomography predict mesorectal fat and presence of vessels posterior to Fascia of Denonvilliers. severity of acute pancreatitis? The purpose of this study is to compare endorectal ultrasound (ERUS) with E. Guler, T. Kose, M. Harman, O. Ozutemiz, N.Z. Elmas; MRI for prediction of involved ARM in LRC. Izmir/TR Material and methods: We identified rectal cancer patients who underwent Purpose: To evaluate the volumes of extrapancreatic inflammation and pan - ERUS over a 5-year period. We reviewed imaging reports and case notes of creas in predicting severity of acute pancreatitis compared with current scor- those with a LRC (within 5cm of anorectal junction) and threatened ARM re- ing systems and laboratory markers. ported on MRI in whom ERUS was undertaken for further evaluation. Histopa- Material and methods: This retrospective study included 30 patients (17 thology was used as a reference standard for the involvement of ARM and women, 13 men; median age 66.5 years) with acute pancreatitis who were sensitivity, specificity, positive predictive value (PPV) and negative predictive examined with computed tomography (CT) in early disease onset. Extrapan- value (NPV) were calculated for ERUS and MRI. creatic inflammation volume (EPIV), pancreatic volume (PV), ratio of EPIV to PV, Results: 24 patients had LRC with threatened ARM reported on MRI. Com- Balthazar score, CT severity index (CTSI), and modified CTSI were calculated. plete data including MRI, ERUS and subsequent histopathology report in 9 Amylase, lipase, C-reactive protein (CRP), and white blood cell (WBC) levels at patients were available, 5 who had naCRT. ERUS agreed with MRI for ARM the time of obtained CT scans were reviewed. Outcome parameters included involvement in 3 cases. 6 patients were appropriately triaged to more exten- the length of hospital stay and development of complications. Spearman’s sive resection (removal of posterior capsule of prostate or posterior vaginal rank correlation and Mann-Whitney U tests were used in the analysis of cor- wall) following confirmation of involved margin by ERUS, and 3 had less exten - relations. sive surgery following confirmation of a clear margin seen at ERUS. Sensitivity, Results: Significant positive correlations were found between EPIV and Balt - specificity, PPV and NPV was 80%, 75%, 80% and 75% respectively for ERUS hazar score, CTSI, and mCTSI (r:0.709, r:0.741, r:0.799, respectively and compared to 40%, 20%, 33% and 25% for MRI. p<0.001 for all). There were no significant correlations between pancreatic vol - Conclusion: ERUS is an accurate technique for complementary assessment ume and scoring systems. Positive correlations between EPIV/PV and 3 scor- of ARM in low rectal cancer. ing systems were depicted (p<0.001). Duration of hospitalization correlated with CTSI and mCTSI (r:0.47 and r:0.546, p<0.05). Significant correlations be - SS 2.10 tween the presence of complications and EPIV, EPIV/PV, CTSI, and mCTSI The diagnostic performance of diffusion-weighted MRI were observed (p<0.05). There were no significant correlations between amyl - and computed tomography in the detection of ase, lipase, CRP, WBC levels and 3 scoring systems. extramural venous invasion in rectal cancer Conclusion: EPIV and EPIV/PV can predict severity of acute pancreatitis. CTSI A. Gursoy Coruh, E. Peker, A. Erden; Ankara/TR and mCTSI calculated in early onset of disease correlate with length of hospital stay and complications. Purpose: To evaluate the diagnostic utility of diffusion-weighted magnetic resonance imaging (DWI-MR) and CT in the detection of extramural venous invasion (EMVI) in rectal cancer. SS 3.2 Material and methods: Out of 358 patients with rectal cancer, totally 58 pa- Changes in radiology workflow in the setting of tients (30 patients with mrEMVI score: 3-4 and ctEMVI score: 2-3 and 28 con- antibiotic treatment for uncomplicated appendicitis trol patients without EMVI) were enrolled in the study. Apparent diffusion coef- D.J. Bowden, M. Sheehan, A. Lee, R. Dunne, A. Hill, M. Lee, ficient (ADC) values of the tumor and EMVI(+) vein were measured. Diameter of M. Morrin; Dublin/IE superior rectal vein (SRV) and inferior mesenteric vein (IMV), distant metastatic Purpose: At a university teaching hospital, a single centre, prospective trial of spread were evaluated on CT. Pathology was accepted as the gold standard. the management of acute uncomplicated appendicitis (AUA), randomising pa- Results: Mean diameters of SRV (4.8±0.9mm vs. 3.6±0.8mm) and IMV tients into conservative treatment with antibiotics versus treatment with sur- (6.9±0.8mm vs. 5.4±0.9mm) were significantly larger (p<0.001) and ADC val - gery was undertaken (COMMA trial NCT#02916134). We analysed the pattern ues of the tumor and the vein were significantly lower (926.4±281.8 s/mm vs. of imaging requested in this group, and compared it with the pattern of imag- 1026.6±246.8 s/mm , p=0.032) in EMVI (+) patients compared to the control ing in patients who had undergone appendicectomy for clinically suspected group. A diameter of 3.95mm for the SRV (sensitivity: 93%, specificity: 66%, appendicitis during the year preceding the trial. The aim is to estimate the im- accuracy: 80%); 5.95 mm for the IMV (sensitivity: 93%, specificity: 71%, ac - pact on radiology workflow in a busy academic centre on transitioning from curacy: 82%) and 0.929 s/mm for ADC value was found as cut-off value by standard surgical treatment of AUA to antibiotic treatment. ROC analysis, for discrimination of EMVI (+) and EMVI (-) patients. When at Material and methods: Patients who underwent appendicectomy before the least two of these 3 criteria are present, sensitivity, specificity values and ac - COMMA trial and all patients enrolled in the COMMA trial were studied (pre- curacy increase (sensitivity: 100%, specificity: 75%, accuracy: 87%). The COMMA Nov 2014–Sept 2015, COMMA Sept 2015 – Dec 2017). The frequen- presence of distant metastases at presentation was significantly more preval - cy of positive imaging and histology was recorded in both groups. ant in EMVI (+) patients (p=0.002). Results: In pre-trial period, 282 appendicectomies were performed. This Conclusion: Measurement of ADC values and SRV-IMV diameters seems to group underwent 63 ultrasound scans (22%), 57 (20%) CTs and 1 MRI (0.3%). be a reliable method in the detection of EMVI in rectal cancer. EMVI (+) patients Negative histology at appendicectomy during the pre-trial period was 23% appear to have a higher risk for distant metastes at diagnosis. (n=64). During the trial period, 259 patients with clinically suspected appendi- citis were assessed. This group underwent 149 ultrasounds (56%), 78 CTs (30%) and 70 MRIs (27%). Of this group, 202 were randomized to antibiotic treatment and 198 to surgery. Negative histology at appendicectomy was 2.5% (n=5). Conclusion: Pre-operative imaging substantially reduces the number of nega- tive appendicectomies. There was a substantial increase in radiology tests ordered in the transition from surgical treatment of AUA to antibiotic treatment which will have considerable implications for radiology workflow and will re - quire significantly increased resources to be diverted to radiology in order to be implemented. S663 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 3.3 SS 3.5 Can CT accurately predict the presence of a single Is unenhanced CT-scan enough for the diagnosis and adhesive band in adhesive small bowel obstruction? the management of elderly patients with acute J. Sammon, J. Byrne, A. Okrainec, A. Hanbidge; abdominal pain in emergency department? Toronto, ON/CA M. Barat, A. Kiani, P. Calame, M. Lagadec, M. Zappa, V. Vilgrain, M. Ronot; Clichy/FR Purpose: Small bowel obstruction (SBO) is a common surgical emergency, with adhesions being the most frequent etiology (65-80%). Laparoscopic man- Purpose: To compare the accuracy and reproducibility of unenhanced and agement has become an established technique, with successful outcomes contrast-enhanced CT-scan for the diagnosis of patients over 75 years old higher in cases of single adhesive band (SB-SBO). Most patients have CT admitted in emergency department (ED) for acute abdominal pain (AAP). prior to surgery, however, the presence of a single adhesive band is usually not Material and methods: 208 consecutive patients over 75 years old consulting commented on in reports. the ED for AAP with a CT-scan with and without enhancement were retrospec- Material and methods: Patients undergoing laparoscopic surgery for adhe- tively included. Three readers read both exams in two different readings: 1) sive SBO 2006–2014 were eligible. 110 patients were identified. 14 patients unenhanced images only, 2) full exam. Sensitivity and specificity for diagnoses were excluded: 12 CT >3 days pre-surgery and 2 no CT features of SBO. Two were compared to the gold standard, defined as the diagnosis obtained after abdominal radiologists independently reviewed the CTs. A single band was a complete clinico-biological and radiological evaluation. Intra and inter-reader recorded for patients who had a single transition point or had two closely re- agreement were calculated for each reader using the kappa test and the inter- lated transition points. Other factors including the degree, presence of a class correlation. Sub-group analyses were performed for patients with an or- closed loop, free fluid, decreased perfusion, mesenteric oedema, bowel wall ganic disease as final diagnosis and for patients requiring an interventional thickening and pneumatosis. treatment. Results: 58 patients successfully underwent laparoscopic surgery, with 38 pa- Results: Diagnostic accuracy ranged from 64% to 68% without enhancement, tients requiring conversion to laparotomy. 33/58 laparoscopic patients had and from 68% to 71% after contrast enhancement. Contrast-enhanced imag- SB-SBO and 9/38 converted patients had SB-SBO. Inter-observer variability es did not significantly improve the diagnosis accuracy (P = 0.973 – 0.979). CT between the two readers for SB-SBO on CT was very good with κ=0.739 (95% scan corrected the diagnosis proposed by the ED physician in 59.1% (range CI: 0.579-0.899) and κ=0.807 (95% CI: 0.687-0.927) with the surgical result. 58.1-60.0%) and 61.2% (range 57.6-65.5%) of the patients before and after Reader 1 accurately assessed the presence of a SB-SBO in 40/42 cases (95% contrast injection (p>0.05). Intra-observer agreement between readings was sensitivity {95% CI: 0.888-1.017}) and Reader 2 in 37/42 cases (88% sensitiv- moderate to substantial (K=0.513-0.711). Inter-reader agreement was sub- ity {95% CI: 0.783-0.979}). stantial for both unenhanced (k=0.745-0.789) and full exam readings (k=0.745- Conclusion: CT can predict the presence of a single adhesive band and pre- 0.799). These results were similar in sub-group analysis. dict more complex cases, helping the surgical team triage patients appropri- Conclusion: Unenhanced CT-scan alone is accurate and reproducible enough ately to a laparoscopic approach, improving patient morbidity and length of for the diagnosis of patients >75 years old presenting with an AAP. hospital stay. SS 3.6 SS 3.4 Clinical utility of functional MRCP in the diagnosis of Increased unenhanced bowel-wall attenuation: a acute cholecystitis specific sign of bowel necrosis in closed-loop small- S. Ramanathan, D. Kumar, M. Heidous, A.A. Faki; Doha/QA bowel obstruction Purpose: To assess the feasibility and diagnostic accuracy of functional 1 2 1 1 C. Rondenet , I. Millet , L. Corno , I. Boulay-Coletta , MRCP (fMRCP) with hepatobiliary contrast in the diagnosis of acute cholecys- 2 1 1 2 P. Taourel , M. Zins ; Paris/FR, Montpellier/FR titis. Purpose: To identify computed tomography (CT) findings associated with Material and methods: 150 patients who underwent gadolinium ethoxybenzyl bowel necrosis in patients with surgically confirmed strangulating closed-loop diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) enhanced MR imaging small-bowel obstruction (CL-SBO) due to adhesions or internal hernia. for various clinical indications were included. Study group comprised 60 pa- Material and methods: This retrospective study was approved by our institu- tients who presented with acute right upper quadrant pain and equivocal tional review board, and informed consent was waived. To identify CT signs of physical examination and/or ultrasound findings for acute cholecystitis. 90 bowel necrosis, two gastrointestinal radiologists performed blinded, inde- patients who came for other indications were considered as control group. pendent, retrospective reviews of 41 CT studies from consecutive patients Gallbladder was assessed for morphological features of acute cholecystitis. who had CL-SBO due to adhesions or internal hernias and who underwent Functional MRCP was evaluated for timing of contrast appearance in the bil- surgery within 48h. Based on surgical and pathological findings, patients were iary confluence, common bile duct (CBD), contrast reflux in to gall bladder classified as having reversible ischemia or histologically documented necrosis. (GB)/cystic duct and excretion into duodenum. Univariate statistical analyses were performed to assess associations between Results: In the control group, contrast reflux into GB/cystic duct occurred in CT signs and bowel necrosis. K statistics were computed to assess interob- 85/90 patients (94.4%) at 15 ± 4 minutes (range: 5-30 min). 5/90 patients who server agreement. did not show reflux demonstrated no extrahepatic biliary excretion due to poor Results: We included 25 (61%) women and 16 (39%) men with a median age hepatocellular uptake for other reasons. In the study group, 35/60 (58%) pa- of 79 years. Bowel necrosis was found in 25/41 (61%) patients and ischemic tients demonstrated reflux within 30 min. All these patients improved on clini - but viable bowel in 16/41 (39%) patients. Increased unenhanced bowel-wall cal and US follow up without acute cholecystitis. 25/60 patients did not dem- attenuation was the only CT finding significantly associated with bowel necro - onstrate contrast reflux till 1 hour delayed image. Acute cholecystitis was sis (P=0.0002). This sign had 58% (95%CI, 37-78) sensitivity and 100% (95% proven in 21/25 patients (84%) by cholecystectomy and development of con- CI, 79-100) specificity for necrosis. Interobserver agreement was fair (0.59; vincing morphological features of acute cholecystitis. 4/25 (16%) patients did 95%CI, 0.37-0.82). not demonstrate contrast reflux. Conclusion: Increased unenhanced bowel-wall attenuation is specific for Conclusion: fMRCP is highly sensitive (96%) and specific (94%) in the diag - bowel necrosis and should lead to prompt surgery for bowel resection. nosis of acute cholecystitis, provided there is optimal extrahepatic biliary ex- cretion. It is particularly helpful in cases of acute cholecystitis where the clini- cal and sonographic features are inconclusive. S664 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 3.7 SS 3.9 Role of imaging in the diagnosis of acute uncomplicated Emergency CT evaluation of pancreatic surgery appendicitis in an adult population in a large prospective complications in relationship to preoperative rectal/ clinical trial pharyngeal swabs status A. Lee, M. Sheehan, D. Bowden, P. O‘Leary, A. Hill, M. Lee, A. Fighera, G. Cardano, A.J. Cybulski, L. Bertuzzo, R. Dunne, M. Morrin; Dublin/IE G. Marchegiani, G.A. Zamboni, G. Mansueto; Verona/IT Purpose: Acute appendicitis is the most common cause of an acute abdomen Purpose: Multiresistant bacteria are becoming more and more common, and in adults. The purpose of this study is to assess the use of imaging in the diag- represent a cause of significant increase in morbidity and mortality in surgical nosis of acute uncomplicated appendicitis (AUA). patients. Our purpose was to analyze a series of patients who underwent Material and methods: Over an 18-month period, a prospective trial of pa- emergency CT for suspected abdominal complications after pancreatic sur- tients who met inclusion criteria for AUA were randomised into two treatment gery, correlating the findings with the preoperative rectal/pharyngeal swabs. limbs based on radiological findings on ultrasound+/-CT or MRI: conservative Material and methods: We searched our picture archiving and communica- management with antibiotic therapy or surgical appendicectomy. CT or MRI tion system (PACS) system for the first emergency postoperative CT scan for were performed if US was inconclusive. suspected abdominal complications after pancreatic surgery performed in our Results: 117 patients that were suspected of having AUA, met the inclusion institution in 1 year, thus obtaining 89 CT scans. Twenty-one patients had criteria and were randomised. 38.5% (N=45/117) had ultrasound alone prior to positive preoperative rectal or pharyngeal swabs, and 68 negative swabs. We randomisation. 19.6% (N=23/117) had MRI and 11.11% (N=13/117) had CT reviewed the scans to assess for the presence and type of complications. following a negative or inconclusive US. 38.5% were randomised after US Results: Among the 89 patients who underwent emergency CT, 21 (23.6%) alone, of which 53.33% (N=24/45) were randomised to the surgical arm. had positive swabs and 68 (66.4%) negative swabs. Patients with positive 79.16% (N= 19/24) had an appendicectomy with a pathologically proven AUA. swabs underwent CT on average 6.6 days after surgery, compared to 8.9 days Of note, negative histology rate was very low at appendicectomy 2.5% (N=5). for patients with negative swabs (p= 0.0021). The prevalence of patients with Discordant cases included 3 pathologically normal appendices, 1 of which had positive swabs was higher among the 28 patients with CT signs of bleeding: a faecolith at surgery; 1 gangrenous appendix (complicated appendicitis); and 28.6% patients with bleeding had positive swabs. For fistulas and collections, 1 granulomatous appendix (thickened but not inflamed). 46.6% (N=21/45) of the proportion of patients with positive or negative swabs was similar to that of whom had US alone were randomised to the conservative arm. Of these, the population who underwent these emergency CT scans. 90.47% (N=19/21) were successfully managed conservatively. 2 patients re- Conclusion: Patients who undergo pancreatic surgery with preoperative posi- turned for an appendicectomy. 1 patient had pathologically proven AUA and 1 tive swabs, undergo abdominal CT for suspected complications on average had a histologically normal appendix. 2.3 days earlier than patients with negative swabs. The prevalence of patients Conclusion: Traditionally, patients suspected of having AUA have been treat- with positive swabs is increased in patients with a CT positive for bleeding. ed with surgical resection. However, pre-operative imaging substantially re- duces the number of unnecessary negative appendicectomies, in addition to SS 3.10 identifying those patients who will respond to a conservative treatment with Computed tomography for acute abdominal antibiotics. haemorrhage: impact upon emergency embolisation A.C. O‘Brien, G.M. Healy, N. Rutledge, A. Patil, SS 3.8 J.W. McCann, C.P. Cantwell; Dublin/IE Acute jejunoileal diverticulitis: multicenter descriptive Purpose: To assess the utility of computed tomography (CT) prior to attempt- study of 33 patients ed embolisation of acute abdominal haemorrhage. 1 2 1 3 3 P. Lebert , I. Millet , O. Ernst , I. Boulay-Coletta , L. Corno , Material and methods: All patients who underwent attempted emergency 2 3 1 2 3 P. Taourel , M. Zins ; Lille/FR, Montpellier/FR, Paris/FR embolisation for acute abdominal haemorrhage at our institution between Purpose: Acute jejunoileal diverticulitis is a very rare and potentially serious 01/01/2010 and 31/12/2016 were included. Obstetric, variceal bleeds and disease involving mostly the elderly. The diagnosis is based on imagery but haemodynamically stable patients were excluded. 138 patients underwent 154 remains unrecognized. The purpose of this study is to describe the clinical and procedures, of which 131 were preceded by CT. Of the 154 procedures, suc- computed tomography (CT) features and the outcomes of acute jejunoileal cessful treatment was performed for active bleeding (74 cases), pseudoaneu- diverticulitis. rysm (26), AV shunt (6), irregular/truncated artery (10) and tumour (3). Empiric Material and methods: Cases of acute jejunoileal diverticulitis managed at treatment was performed in 20 cases. three French hospitals were identified retrospectively from 2005 through 2015. Results: Performing a CT before attempted emergency embolisation was as- The final diagnosis relied either on a clinical and radiological data review by a sociated with a significantly higher chance of successful treatment compared panel of experts, or on the surgical findings. Demographic, clinical, laboratory, to those with no CT (81% vs 59% respectively, p=0.023). For those who under- and 18-month outcome data were collected. CT scans were reviewed by hav- went CT, an acute vascular finding (active bleed, pseudoaneurysm, AV shunt, ing two radiologists reach a consensus about the inflammatory diverticulum, irregular/truncated artery) was associated with a higher chance of successful evidence of complications, and presence of other bowel diverticula. treatment compared to those with no acute vascular finding (85% vs 52%, Results: We identified 33 cases in 33 patients with a median age of 78 years, p=0.002). Patients with no acute vascular findings on CT (n=18), demonstrated including 30 (86%) in whom an inflammatory diverticulum was identified, at the active bleeding (5), pseudoaneurysm (2) and irregular/truncated artery (3) on jejunum (n=26, 87%) or ileum (n=4, 13%). Extra-intestinal gas was seen in 10 formal angiography. (30%) and fluid in 11 (33%) patients. Other small-bowel diverticula were visible Conclusion: Performing CT prior to emergency embolisation for acute ab- in all 33 patients. The diverticulitis was mild and resolved with non-operative dominal haemorrhage increases the probability of successful treatment. An treatment in 22 (67%) patients and severe in the remaining 11 (33%) patients, acute vascular finding on CT further increases the probability of success. How - of whom 8 required emergency surgery. ever, the absence of acute vascular findings should not exclude consideration Conclusion: Acute jejunoileal diverticulitis is a rare and usually non-serious for embolisation in the acute patient. condition that chiefly involves the jejunum. A detailed CT assessment may al - low non-operative treatment. S665 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 Wicklow Hall 2A SS 4.2 Diagnostic accuracy of standardised uptake value Scientific Session SS 4 (SUVmax) in predicting malignancy of supraclavicular Abdominal oncology lymph nodes from primary oesophageal cancer C. O‘Rourke, I. Welaratne, S. Cournane, L. McLoughlin, J.V. Reynolds, C. Johnston, N. Sheehy; Dublin/IE SS 4.1 Diagnostic value of dynamic perfusion MRI imaging in Purpose: The purpose of this study was to determine the diagnostic accuracy patients with local advanced rectal cancer in the and optimum cut-off value of SUVmax on positron emission tomography (PET) assessment of chemo-radiation treatment: relation to to predict malignancy of supraclavicular lymph nodes (SCLNs) in patients with tumor regression grade at histology oesophageal carcinoma. S.G. Drago, A. Pecorelli, C. Talei Franzesi, M. Cressoni, Material and methods: All diagnosed cases of oesophageal cancer were ret- S. Sironi, D. Ippolito; Monza/IT rospectively reviewed (2012-2016). Patients that had a confirmed diagnosis of oesophageal cancer with avid SCLNs on staging PET were included in the Purpose: To investigate the value of dynamic contrast-enhanced perfusion- study. 33 SCLNs that subsequently underwent ultrasound guided biopsy for MRI parameters in the evaluation of the response to chemo-radiation therapy staging were analysed. The maximum uptake values (SUVmax) of the SCLNs in patients with local advanced rectal cancer in comparison with histology. and primary tumours were measured. A receiver operating characteristic Material and methods: Fourteen patients with biopsy proven rectal adeno- (ROC) analysis was performed to determine the optimum cut off of SUVmax in carcinoma who underwent dynamic contrast-enhanced MR performed on predicting malignancy. 1.5T MRI system (Achieva, Philips), before (MR1) and after chemo-radiation Results: 24/33 PET-detected SCLNs were malignant. ROC analysis identified therapy (MR2), were enrolled in this study. The protocol included T1 gadolini- the best nodal SUVmax cut-off to be 3.0. The diagnostic accuracy of PET was um-enhanced THRIVE sequences acquired on axial planes. A dedicated work- 76.0% (sensitivity = 78.9%, specificity = 66.6%). For SCLNs with SUVmax > station was used to generate color permeability maps; region of Interest (ROI) 3.0, PET showed a positive predictor value (PPV) of 88.2%; for SCLNs < 3.0, was manually drawn on tumor tissue and normal rectal wall, hence the follow- PET showed a negative predictor value of 50%. The median SUVmax of path- ing parameters were calculated and statistically analyzed: Relative venous ologically negative and positive nodes were 2.8 (range 1.8 - 6.0) and 5.3 (range enhancement (RVE), maximum enhancement (ME) and time to peak (TTP). 1.9 -13.4). The median primary tumour SUVmax was 13.8 (range 3.7 - 30.0). Perfusion parameters were related to pathologic tumor regression grade (Man- The SUVmax of metastatic lymph nodes were significantly higher than those of dard’s criteria; TRG1=complete regression, TRG5=no regression). benign lesions (p < 0.05). Results: Five tumors (36%) showed complete or subtotal regression (TRG1-2) Conclusion: Our study revealed an accuracy rate of 76% for PET detected at histology and classified as responders; 9 tumors (64%) were classified as SCLNs in patients with oesophageal carcinoma. For SCLNs with SUVmax > non-responders (TRG3-5). Perfusion MRI parameters were significantly higher 3.0, PET had a high PPV (88%), which can minimize the need for further diag- in the tumor tissue than healthy tissue in MR1 and MR2 (p<0,05). At baseline nostic tests. (MR1), no significant difference in perfusion parameters was found between responders and non-responders. After chemo-radiation therapy, at MR2, re- sponders showed significantly (p< 0.05) lower perfusion values (RVE(%)76±25; SS 4.3 ME(%):96±41;TTP(sec):13±26) compared to non-responders (RVE(%):153±50; Dedicated DW-MR imaging as a selection tool for ME(%):169±42;TTP(sec):38±68). Moreover, in responders, perfusion values hyperthermic intraperitoneal chemotherapy in patients decreased significantly at MR2 (RVE(%)76±25; ME(%):96±41;TTP(sec):13±26) with peritoneal carcinomatosis from colorectal origin compared to MR1 (RVE(%):118±15;ME(%):120±18;TTP(sec): 14±24) (p<0,05). I. Van‘t Sant, M. Engbersen, H. Van Eden, Furthermore, in non-responders, there was no significant difference between D.M.J. Lambregts, N.F.M. Kok, A.G.J. Aalbers, perfusion values at MR1 and MR2. R.G.H. Beets-Tan, M.J. Lahaye; Amsterdam/NL Conclusion: Dynamic contrast perfusion-MRI analysis represents a comple- Purpose: Combined cytoreductive surgery (CRS)/hyperthermic intraperitoneal mentary diagnostic tool for identifying vascularity characteristics of tumor tis- chemotherapy (HIPEC) treatment gives a subset of patients with peritoneal sue in local advanced rectal cancer, useful in the assessment of treatment re- carcinomatosis (PC) from colorectal origin (CRPC) with a limited and resecta- sponse. ble peritoneal tumor load and prospect of long-term survival or even cure. A non-invasive imaging method to accurately select potential CRS/HIPEC can- didates preoperatively is highly demanded. We compared the peritoneal can- cer index (PCI) estimated preoperatively by diffusion-weighted MR imaging (DW-MRI) with PCI found at surgery to assess whether DW-MRI can be used to select CRS/HIPEC candidates. Material and methods: Forty-nine consecutive patients (M/F=23/29) with confirmed or suspected PC from histologically proven colorectal origin were included. All patients underwent preoperative dedicated whole-body DW-MRI before exploratory laparoscopy or CRS/HIPEC. PCI was prospectively as- sessed by two independent radiologists on MR-imaging and was compared to PCI found at surgery (reference standard). Based on PCI patients were catego- rized as operable (PCI 0-20) or inoperable (PCI 21-39). Results: Mean PCI at surgery was 12.2 (range 0-34). Mean radiological PCI for reader 1 was 10.9 (range 2-33) and 9.2 (range 0-32) for reader 2. Radiologist 1 and 2 staged respectively 46/49 and 44/49 patients correctly (accuracy 94%/90%). Both radiologists detected all operable patients with a PCI<21 at surgery (sens=100%). Respectively 3 and 5 patients were understaged by DW-MRI. No patients were overstaged. The intraclass correlation between the radiologists was excellent (ICC=0.91, 95% CI: 0.758-0.959). Conclusion: In a large cohort, DW-MRI seems to be an accurate and robust selection tool to noninvasively select CRPC patients in which CRS/HIPEC is feasible. Further multicenter studies must be performed to see whether DW- MRI may replace surgical PCI staging. S666 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 4.4 SS 4.6 Radiological considerations in multiple site CT-liver-perfusion dose reduction protocols for cytoreductive surgery for metastatic colorectal cancer: detection of hepatocellular carcinomas: is it achievable? a cohort study A.A. Hatzidakis, G. Kalarakis, K. Perisinakis, N. Kosidekakis, A.D. Stirling, A.N. Murphy, C.S. McQuade, J. Mulsow, N. Kokkinos, A. Karantanas; Heraklion/GR C. Shields, H. Fenlon, C.G. Cronin; Dublin/IE Purpose: To compare the diagnostic efficiency of standard “35-passes” CT- Purpose: Cytoreductive surgery (CRS) can improve survival in selected pa- liver-perfusion (CTLP) acquisition protocol versus two “18-passes” CTLP-pro- tients with metastatic colorectal cancer. These are complex, major surgeries tocols for hepatocellular carcinoma (HCC) detection. that should be undertaken in specialist centres. The aim of this study is to Material and methods: Fourteen patients (13 male, age 57-87) with 32 con- evaluate outcomes relevant to radiological practice and resource management firmed HCCs (median diameter 25mm, 11-125mm) underwent CTLP using a in patients undergoing multiple site CRS for metastatic colorectal cancer. high temporal resolution acquisition protocol with 35 passes, pass-duration Material and methods: Patients who underwent multiple site CRS were iden- 1.7 sec, and total duration 59 sec. Eleven different perfusion maps were gener- tified from a prospectively maintained database of all patients with metastatic ated. Regions of interest (ROIs) were positioned on non-tumorous parenchyma colorectal cancer referred to a national centre for the management of colorec- and on HCCs. Maps were also retrospectively generated using the 18 first tal cancer and peritoneal malignancy. The number and anatomical location of passes (total acquisition time 31 sec) and the 18 odd numbered passes (total the operative sites, peritoneal carcinomatosis index (PCI) and completeness of acquisition time 59 sec). Receiver operating characteristic (ROC) analysis was cytoreduction were recorded from electronic medical records. The number of employed to evaluate the ability of each map to discriminate HCCs from non- multidisciplinary discussions, the extent of follow-up imaging and the number tumorous liver parenchyma. Comparison of ROC curves was performed to of image-guided interventions were recorded from a national picture archiving evaluate statistical significance of differences in the discriminating efficiency of and communication system (NIMIS PACS). the derived perfusion maps between different acquisition protocols. Results: 72 patients were included over a 3-year period from May 2014-2017. Results: Hepatic arterial blood flow, mean slope of increase, time to peak, The mean number of operative sites was 3 (range 2-7). The most common positive enhancement integral, and hepatic arterial fraction were found to be sites of CRS were: peritoneum and omentum (43%), liver (38%), and female the prevailing parameters regarding HCC discrimination when the “35-passes” reproductive organs (22%). This cohort had extensive imaging requirements protocol was used, with estimated area under ROC curve of 1.000, 0.995, averaging 10.5 investigations (either CT, MRI or positron emission tomogra- 0.993, 0.990, 0.983 respectively, without significant differences in their dis - phy/CT, excluding initial staging) and 4.4 multidisciplinary discussions. The criminating power (p>0.05), followed by Tmax, mean transit time (MTT), per- mean number of image-guided interventions was 2.4 (range 0-16). meability–surface area product (PS), blood flow (BF), blood volume (BV), Conclusion: The radiology input to the treatment paradigm of those undergo- IRFT0. Comparison of ROC curves between “35-passes” and “18-passes” ing CRS for metastatic colon cancer is substantial. Awareness of this will help protocols yielded no significant differences (p>0.05) for any of the studied guide resource management in this complex patient cohort. maps. Conclusion: Dose reduction in CTLP for HCC detection can be safely achieved by reducing the number of passes to 18, since the diagnostic accuracy of the SS 4.5 produced perfusion maps is not downgraded. Role of gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid MRI in characterizing nodules with atypical enhancement on CT: results of a prospective SS 4.7 monocentric study Peri-hilar cholangiocarcinoma preoperative CT and MRI: R. Faletti, M. Gatti, E. Caramia, A.M. Bozzato, A. Ferraris, accuracy and inter-observer agreement 1 1 1 1 S. Fiore, L. Bergamasco, P. Fonio; Turin/IT A. Sarno , A. Contro , A. Ruzzenente , S. Conci , 1 1 2 2 G. Tedesco , V. Ciaravino , R. De Robertis , N. Cardobi , Purpose: To assess the use of gadolinium ethoxybenzyl diethylenetriamine 1 1 2 M. D‘Onofrio ; Verona/IT, Peschiera del Garda/IT pentaacetic acid (Gd-EOB-DTPA) MRI in patients with “atypical” HCC diag- nosed at multiphase CT. Purpose: To evaluate the accuracy of CT and MRI in the preoperative staging Material and methods: 51 cirrhotic patients with at least one “naïve” lesion of peri-hilar cholangiocarcinoma. showing atypical dynamics enhancement on multiphase CT underwent Gd- Material and methods: CT and MRI study of 60 patients (45 males and 15 EOB-DTPA MRI were prospectively enrolled. Images were reviewed by two females, mean age 67 years) with peri-hilar cholangiocarcinoma were retro- radiologists in consensus. The atypical lesions were re-categorized by MRI spectively reviewed by two radiologists. All patients underwent surgical evalu- into: group A) typical HCC; group B) hypervascular without washout and hy- ation. The review of imaging data was performed blinded to surgical and pointense in the hepatobiliary phase (HBP) and group C) hypovascular nod- pathological results and separately. Involvement of biliary systems and vessels ules, hypointense in the HBP, hyperintense on diffusion-weighted imgaing was judged. Sensitivity, specificity and accuracy were calculated with respect (DWI). Patients were classified according to the Barcelona Clinic Liver Cancer to surgical and pathological results. Inter-observer agreement was calculated (BCLC) guidelines prior and after the MRI. Data were analysed using non-par- with K agreement test. ametric tests. Results: 50 patients were resected with curative intent (R0/R1) and 10 patients Results: 15 typical and 75 atypical lesions were found at CT. Twenty-six (66%) underwent explorative/palliative surgical treatment. The right bile ducts were atypical lesions were re-classified into in group A, 8 (21%) in group B and 5 involved in 31 patients, the left in 42 and both were involved in 26. Portal vein (13%) in group C; among the atypical lesion 17 were not confirmed, 12 were and hepatic artery involvement were present respectively in 36,2% and 23,2% not detected and the diagnosis remained uncertain in 7 at MRI. Typical lesions of patients. Lymph-node involvement was observed in 25 patients. The sensi- resulted bigger than atypical lesions (22.3±16.9 vs. 12.3±7.3 mm; p=0.0004) bility, specificity and accuracy of MR and CT were similar. Accuracy for CT was and in particular, a lesion >15mm had a positive likelihood ratio (LR+) of 2.25 76.4%, 79.1% and 55.1% for biliary, vascular and lymph-node involvement. to be typical. BCLC stage was changed after MRI in 61% (31/51) patients. For MR accuracy was 74.0%, 82.2% and 51.6%, respectively. Inter-observer Based on MRI a biopsy was indicated for 7 nodules compared to 43 based on agreement (K value) was higher for MR compared to CT in biliary involvement CT and 75% of group C lesions were confirmed histologically as HCC. (0.57 and 0.41), lower for arterial involvement (0.41 and 0.44) and lymph-node Conclusion: Gd-EOB-DTPA could correctly classify the majority of small and involvement (0.18 and 0.22). The agreement of the two readers showed statis- “atypical” HCC found on CT with high clinical impact on BCLC staging and tically significant differences (p<0.001) for the right lobe. management decision. Conclusion: CT and MRI are accurate in peri-hilar cholangiocarcinoma stag- ing. MRI is superior to judge biliary systems involvement. Inter-observer agree- ment was inferior for the right lobe. S667 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 4.8 SS 4.10 Improved delineation of hepatic metastases from Comparison of gadobutrol and gadofosveset trisodium colorectal cancer using noise-optimized virtual in MRI for rectal cancer 1 2 3 monoenergetic reconstructions from dual-energy R.A.P. Dijkhoff , J. Santinha , S.G. Drago , 1 1 4 computed tomography D.M.J. Lambregts , J. Van Griethuysen , F.C.H. Bakers , 1 1 1 2 1 2 1 J.L. Wichmann , M.H. Albrecht , S.S. Martin , L. Lenga , R.G.H. Beets-Tan , N. Papanikolaou , M. Maas ; 1 1 1 1 1 2 3 4 D. Leithner , B. Bodelle , T.J. Vogl , R. Hammerstingl ; Amsterdam/NL, Lisbon/PT, Monza/IT, Maastricht/NL 1 2 Frankfurt am Main/DE, Essen/DE Purpose: To compare the pharmacokinetics of gadobutrol (GDB; micromo- Purpose: To assess the impact of noise-optimized virtual monoenergetic im- lecular contrast-agent) with the albumin-binding contrast-agent gadofosveset aging (VMI+) reconstructions on quantitative and qualitative image parameters trisodium (GDF) in MRI for staging or rectal cancer using semiquantitative in patients with hepatic metastases from colorectal cancer at abdominal dual- analyses. energy computed tomography (DECT). Material and methods: 12 patients with primary rectal cancer were included. Material and methods: Forty-two patients (29 men; 67.1±12.3 years) with 6 patients underwent dynamic contrast-enhanced-MRI (DCE-MRI) with GDF histopathologically-confirmed hepatic metastases from colorectal cancer un - and were matched with 6 patients who had DCE-MRI with GDB. DCE-MRI derwent triphasic contrast-enhanced abdominal DECT. Images from the por- protocol was the same in all patients. Regions of interests were manually tal-venous phase were post-processed with the standard linear blending tech- placed in the tumour and DCE-MRI was processed with OncoAnalysis Horos nique (M_0.6) as well as with the traditional virtual monoenergetic (VMI) and Plugin (Computational Clinical Imaging Group, Champalimaud). Semiquantita- novel VMI+ algorithms in 10-keV intervals from 40 to 100 keV. Attenuation tive parameters (wash-in-slope, wash-out-slope, area-under-the-curve (AUC), measurements were performed in hepatic lesions, unaffected liver tissue, rec- maximum enhancement (MaxEnh) and time-to-peak (TTP)) based on signal tus abdominis muscle, and adjacent visceral fat to calculate objective signal- intensity time curves were extracted and compared between GDB and GDF. to-noise (SNR) and contrast-to-noise (CNR) ratios. Five-point scales were Using the coefficient of variance (standard deviation/mean) heterogeneity was used to evaluate overall image quality and lesion delineation by three blinded calculated for all semiquantitative parameters. radiologists with varying levels of experience. Results: Contrast-uptake parameters such as mean wash-in-slope (19.86 vs. Results: SNR and CNR (negative values indicating best contrast of hypoat- 22.16, respectively, p=0.690) and mean maximum enhancement (283.72 vs. tenuating lesions) of liver metastases were best in 50-keV VMI+ series (3.5±2.1 286.86, respectively p=0.615) were comparable between patients with GDB and -3.6±2.0), significantly superior to all other reconstructions (all P<0.001; and GDF, with only slight differences between both contrast agents. Also, SNR M_0.6: 2.4±1.5; CNR M_0.6: -2.6±1.8). Qualitative image parameters maximum wash-in-slope (58.71 vs. 57.88, p=0.974) was comparable for both showed highest values for 50-keV VMI+ reconstructions (median 5, respec- contrast agents. The differences between GDB and GDF for maximum/mini- tively; P≤0.023) regarding overall image quality. Qualitative assessment of le- mum/mean washout-slope were remarkably small, given the difference in size sion delineation peaked in 40-keV VMI+ (median 5) and 50-keV VMI+ (median of the agents. GDF led to a more homogeneous maximum enhancement than 4; P=0.067), significantly superior to all other reconstructions (all P<0.001; GDB (0.55 vs. 0.33, respectively, p=0.079). M_0.6: median 3). Conclusion: The albumin-binding contrast agent GDF shows similar contrast- Conclusion: Novel VMI+ reconstructions at 50 keV can substantially increase uptake parameters compared to micromolecular GDB in this small sample of quantitative image quality and improve subjective assessment of image quality rectal tumours. It seems that GDF behaves similarly to a micromolecular agent and lesion delineation of hepatic metastases from colorectal cancer compared shortly after injection, which might be explained by the absence of binding to to standard image reconstruction and traditional VMI. albumin immediately after injection. Possibly, results of DCE-MRI with GDF might be extrapolated to micromolecular agents, regarding wash-in phase. SS 4.9 How are we doing? Radiologist and multi-disciplinary team clinician opinions on the quality of MRI rectal cancer staging reports P.J. Brown, H. Rossington, D.J.M. Tolan; Leeds/UK Purpose: Rectal cancer staging with magnetic resonance imaging (RCS-MRI) allows accurate assessment of tumours. Significant variability exists in the content of reports. Colorectal multi-disciplinary teams (CRC-MDTs) make treatment decisions based on these reports. We evaluated the satisfaction of CRC-MDT members with current standards of radiology reporting to identify areas for improvement. Material and methods: 16 UK CRC-MDTs serving a population over 5 million in Yorkshire, were invited to complete an online questionnaire assessing opin- ions on RCS-MRI report content and clarity. This included questionnaires for RCS-MRI radiologist ‘reporters’ and other CRC-MDT ‘users’ who use the re- ports to plan care (including colorectal surgeons, oncologists, pathologists and nurse specialists). 21 of 42 (50%) ‘reporters’ and 48 ‘users’ completed the survey. All RCS-MRI reporting radiologists were subspecialists in GI imaging. Results: Questionnaires indicated items which were ‘always’ contained in re- ports, with tumour stage, nodal status and CRM status included in more than 80% of cases. However there was lower than expected reporting for tumour distance from the anal verge (81% ‘reporters’ v 65% ‘users’), relationship to peritoneal reflection (29% ‘reporters’ v 46% ‘users’), extramural vascular inva - sion (57% ‘reporters’ v 67% ‘users’) and inter-reporter consistency (76% ‘re- porters’ v 75% ‘users’). There was no statistically significant difference be - tween the opinions of reporters and users. Despite this 92% of ‘users’ were satisfied or very satisfied (on a five-point scale) with RCS-MRI. Conclusion: CRC-MDT users were satisfied with RCS-MRI reports despite inconsistencies for reporting critical variables. CRC-MDT user satisfaction of reports may not act as a sufficient stimulus to improve standards of RCS-MRI reporting. S668 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 Wicklow Hall 2B SS 5.2 MRI for response assessment after neoadjuvant Scientific Session SS 5 chemoradiotherapy in oesophageal cancer: added value Updates in upper and lower GI tract disor- of diffusion-weighted imaging S.E. Vollenbrock, F.E.M. Voncken, D.M.J. Lambregts, ders M. Maas, J.M. Van Dieren, L.C. Ter Beek, B.M.P. Aleman, R.G.H. Beets-Tan, A. Bartels-Rutten; Amsterdam/NL SS 5.1 Purpose: High accuracy is needed to select oesophageal cancer patients with Post chemoradiotherapy assessment of anal carcinoma a complete response (CR) after neoadjuvant chemoradiation (nCRT) for organ- using MRI: comparison with positron emission preserving treatment instead of oesophagectomy. Performance of T2-weight- tomography in biopsy proven patients ed MRI (T2W-MRI) and additional diffusion-weighted MRI (T2W+DW-MRI) was R. Faletti, M. Gatti, V. Carrozzo, L. Bergamasco, P. Fonio; determined. Turin/IT Material and methods: Thirty-seven patients with locally advanced oesopha- Purpose: To assess the use of MRI in local response assessment and detec- geal cancer underwent MRI (1.5 Tesla; T2W-MRI and DW-MRI, b-val- tion of early local relapse after chemoradiotherapy (CRT) in patients with anal ues=0,200,800 s/mm ) before and after nCRT with a maximum of 21 days be- squamous cell carcinoma (AC). tween MRI and surgery. Three radiologists initially scored T2W-MRI only. Material and methods: Eight patients with histologically proven AC who com- Thereafter they rescored T2W with DW-MRI. A 5-point score was used (1=def- pleted pre-(t0), post-CRT 4-week(t1) and 20-week(t2) MRIs and pre-(t0) and inite CR, 3=inconclusive, 5=definite residual tumour). Histopathology after oe - post-CRT 20-week(t2) positron emission tomography (PET) scan were pro- sophagectomy represented the reference standard (Mandard tumour regres- spectively included. Images were reviewed by two radiologists in consensus, sion grade 1=CR, 2-5=residual tumour). Area under the receiver operating blinded to clinical outcome: tumor size (D-mm), apparent diffusion coefficient characteristic (AUROC), sensitivity and specificity were calculated considering (ADC-mm /s), time to peak (TTP-sec) and standard uptake value (SUV) were MRI scores 3-5 as tumour-positive. recorded. Lesion biopsy performed at 6-month was the gold standard test for Results: Three out of 37 patients were excluded due to poor image quality. evaluation of response to treatments. Data were analyzed using chi-squared Seven (21%) of the remaining 34 patients achieved a CR. AUROCs were 0.57, test, Wilcoxon signed-rank test and Pearson’s correlation coefficient. 0.66 and 0.66 for T2W-MRI only and 0.75, 0.70 and 0.71 for T2W+DW-MRI Results: All patients were classified as responders. ADC, TTP and D were (p=0.09, 0.32, 0.70). Sensitivity for detecting residual tumour was 100%, 89% significantly different from t0 to t1 (0.8±0.2vs.1.2±0.2; p=0.002; and 89% for T2W-MRI only and 93%, 96% and 85% for T2W+DW-MRI. Spec- 95±55vs.191±59;p=0.03 and 31±14.5vs.18±17;p=0.002) and from t0 to t2 ificity improved from 14%, 43% and 43% on T2W-MRI to 57%, 43% and 57% (0.8±0.2vs.1.4±0.2;p=0.001, 95±55vs.214±67;p=0.006 and for T2W+DW-MRI. 31±14.5vs.10.5±10;p=0.001). From t1 to t2 ADC and TTP were not different Conclusion: T2W+DW-MRI is promising for response assessment after nCRT (p=0.12 and p=0.50) unlike D (p=0.04). A difference from t0 to t1 of 20% in in oesophageal cancer patients. Sensitivity is high, indicating that the risk of ADC (increase), TTP (reduction) and D (reduction) was identified as forecast of missing residual tumour is minimal. Specificity is lacking, indicating that the responder. SUV was significantly different from t0 to t2 (9.9±4vs.1.2±2.1;p=0.02). degree of residual tumour is overestimated. A multimodality approach with ADC was strongly inversely correlated with SUV values (r= -0.74). endoscopy, endosonography and positron emission tomography (PET)-CT Conclusion: D, ADC and TTP summarized in “the-rule-of-20%” seemed to be may further improve response assessment. a good marker of early (4-week) response to treatment. The correlation ADC/ SUV suggested an association between tumor cellularity and metabolic activ- SS 5.3 ity. ADC had a stronger relationship with outcome and may be more useful for Prognostic value of positron emission tomography predicting the prognosis of AC than SUV. (PET)-CT in oesophageal cancer patients undergoing oesophagectomy E. O‘Malley, C. Waters, C. Collins, J. Bruzzi; Galway/IE Purpose: Above the age of 50 oesophageal cancer incidence is 30-40/100,000 in the Irish Health Service Executive Western area. Stage I disease has 60% overall survival at 5 years, stage II 30%, stage III 20% and stage IV less than 5%. Accurate staging is important when selecting surgical candidates and multiple modalities are used including CT, endoscopic ultrasound and PET CT. PET CT may provide extra prognostic value which is not used in clinical prac- tice. Conflicting data exist about the usefulness of maximum standardised uptake value (SUV ) and other measurements such as metabolic tumour vol- max ume (MTV) and total lesion glycolysis (TLG). This study analyses PET CT find - ings and compares tumour measurements including MTV and TLG with the post-surgical mandard response score, 5-year survival and histological stag- ing. Material and methods: This retrospective analysis looked at PET CT studies from 2010 to 2016. Patient demographics, survival time, histopathology stag- ing and PET CT measurements were collected. PET CT measurements includ- ed SUV , SUV , SUV MTV and TLG. max peak mean Results: We identified 64 patients who underwent oesophagectomy. Median age at surgery was 64 (39-78). Histopathology was adenocarcinoma 72% and squamous cell carcinoma 28%. Mean MTV was 47.9 ml, mean SUV 9.9, max mean SUV 7.95, mean SUV 6.23 and mean TLG 207.79. Mean 5-year peak mean survival was 1.5 years. There was no significant correlation between PET CT measurements and survival, mandard score or histological staging. Conclusion: FDG avid measurements alone are not a useful prognostic factor in oesophageal cancer patients undergoing surgery. S669 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 5.4 SS 5.6 Restaging of resectable esophageal cancer using An investigation of the association between sarcopenia positron emission tomography (PET)/CT after and post-operative morbidity and mortality in patients neoadjuvant treatment: combining the best of both with gastric cancer using CT worlds S. O‘Brien, M. Twomey, F. Moloney, R.G. Kavanagh, D. Tamandl, B. Füger, A. Haug, A. Ba-Ssalamah; Vienna/AT M. Maher, O.J. O‘Connor, C. O‘Suilleabhain; Cork/IE Purpose: To assess the feasibility and diagnostic performance of PET/CT in Purpose: Surgical resection for gastric adenocarcinoma is associated with the restaging of esophageal cancer after neoadjuvant therapy. significant post-operative morbidity and mortality. The aim of this study was to Material and methods: We included 88 esophageal cancer patients without assess the prognostic significance of sarcopenia in patients undergoing cura - distant metastases, who were resected after neoadjuvant treatment. PET/CT tive resection for gastric adenocarcinoma with respect to post-operative mor- with a diagnostic, contrast-enhanced CT (CE-CT) scan, was performed before bidity, disease-free and overall survival. resection and the radiological staging was compared to postoperative histo- Material and methods: A retrospective analysis of a cohort of consecutive pathology as the reference standard. Locoregional stage (T, N-stage) was re- patients who underwent surgical resection for gastric adenocarcinoma be- trieved from the CT scan, whereas PET analysis consisted of measurement of tween 2008 and 2014 was conducted. Patient demographics, radiological pa- standardized uptake value (SUV) mean, SUV max, Tumor diameter, metabolic rameters and pathological data were collected. Osirix was used to measure tumor volume (MTV) and total lesion glycolysis (TLG). Sensitivity, specificity skeletal muscle area at a set level, which was normalized for height to calcu- and accuracy of the CT staging in reference to pathology were calculated. PET late skeletal muscle index. Sarcopenia entailed a skeletal muscle index of less 2 2 2 2 parameters were compared using analysis of variance (ANOVA). than 52.4cm /m for men and 38.5cm /m for women. Statistical analysis was Results: Of the 88 patients amenable for analysis, eleven patients (13%) had performed using the statistical package for the social sciences (SPSS). Re- received combination radiochemotherapy, whereas 75 patients (85%) had gression analysis was used to identify predictors of outcomes. neoadjuvant chemotherapy either in combination or monotherapy with various Results: Fifty-six patients (41 male, 15 female, mean age 68.4±11.9 years) met agents. For overall T-staging, true positive, false positive and false negative the inclusion criteria. Thirty-six percent (20/56) of patients were sarcopenic rates for CT vs. PET/CT (qualitatively) were: 59.1% vs. 68.1%, 25% vs 15.9% pre-operatively. Both sarcopenic and non-sarcopenic patient groups were and 15.9% vs. 15.9%, leading to an accuracy of 57.6% vs. 68.2%. For T3+4 equally matched with the exception of weight and body mass index (p=0.036 stages alone, CT had a better diagnostic performance (sensitivity 85.7, speci- and p=0.001). Sarcopenia was associated with a decreased overall survival ficity 81.8%). Volumetric PET parameters, but neither SUVmax nor SUVmean (Log Rank=0.003) and was an adverse prognostic predictor of overall survival correlated with T- stage: MTV (p=.011), TLG (p=0.35), PET-Diameter (p=0.001). on multivariable analysis (hazard ratio=10.915, p=0.001). Sarcopenia was a Likewise, there was an association to American joint committee on cancer predictor of in-hospital serious complications on multivariate analysis (odds (AJCC) stage: MTV (p=.015), TLG (p=0.37), PET-Diameter (p=0.002). ratio=3.508, p=0.042). Conclusion: In restaging of esophageal cancer after neoadjuvant therapy, Conclusion: In patients undergoing curative resection for gastric cancer, there both CE-CT and PET provide independent information on the post-therapeutic is a statistically significant association between sarcopenia and both de - locoregional tumor stage. creased overall survival and serious post-operative complications. The meas- urement and reporting of skeletal muscle index on pre-operative CT should be considered for patient preparation purposes. SS 5.5 More frequent follow-up CT examinations after surgical resection than endoscopic resection of early gastric SS 5.7 cancers: are they really useful? Prediction of tumor response to chemotherapy using S.E. Jung, M.H. Choi; Seoul/KR perfusion CT in patients with unresectable advanced gastric cancer Purpose: A recommended interval to follow-up CT or endoscopy after treat- 1 1 2 1 2 D.H. Lee , S.H. Kim , S.M. Lee ; Seoul/KR, AnYang/KR ment is not provided in the national comprehensive cancer network (NCCN) guidelines. No study has analyzed the role and interval of CT after both surgi- Purpose: To evaluate whether parameters obtained from perfusion CT (PCT) cal and endoscopic resection for gastric cancer. To evaluate the utility and can predict treatment response after palliative chemotherapy in patients with timing of CT in the detection of recurrent tumor after surgical and endoscopic unresectable advanced gastric cancer (AGC). resection (endoscopic submucosal dissection) for early gastric cancer (EGC). Material and methods: This prospective study was approved by the institu- Material and methods: A total of 670 patients after surgical (n = 535) and tional review board and informed consents were obtained. Twenty-one pa- endoscopic resection (n = 135) for EGC between 2007 and 2009 were enrolled. tients with unresectable AGC (M:F=13:8; mean age, 53.7 years old) were in- Mean numbers of CT and endoscopy between both treatment groups were cluded. We performed PCT and palliative chemotherapy in all 21 patients. compared. The mean and cumulative dose length product (DLP) of CTs were After palliative chemotherapy, treatment response was assessed for each pa- calculated. Recurrence-free survival and overall survival were evaluated using tient using RECIST ver. 1.1 and patients who achieved complete or partial re- a Kaplan-Meier method. sponse were classified into responder group. Relationship between tumor re - Results: The mean interval of CT was significantly shorter and the mean num - sponse to palliative chemotherapy and PCT parameters was evaluated using ber of CTs was significantly larger in the surgical resection group than in the the Mann-Whitney test and receiver operating characteristic analysis. endoscopic resection group. All 34 gastric recurrences were diagnosed by Results: After chemotherapy, 6 patients showing partial response were classi- endoscopy. Six extragastric recurrences were detected by CT out of total 5417 fied into the responder group and the remaining 15 patients into the non-re - CT examinations. The average cumulative DLP was significantly higher in the sponder group. Permeability surface value was significantly different between surgical than in the endoscopic resection group (P = 0.004). two groups (51.0 ml/100g/min in responder group versus 23.4 ml/100g/min in Conclusion: Follow-up CT was more frequently performed in patients after non-responder group, P=0.002). Other PCT parameters were not significantly surgical resection than in those after endoscopic resection of EGCs even different between two groups. The area under the curve to predict responders though CT showed limited usefulness for detecting recurrent tumors after ei- was 0.911 (95% confidence interval, 0.787-0.990; P=0.004) for permeability ther treatment methods. Since there was no extragastric recurrence in patients surface value and the sensitivity of 100% (6/6) and specificity of 80% (12/15) with mucosal cancer after resection, routine use of CT should be avoided after were achieved when a cut-off permeability surface value was set at 29.7 treatment of mucosal cancer. ml/100g/min. Conclusion: Obtaining perfusion parameters from PCT was feasible for pa- tients with unresectable AGC and can aid the prediction of treatment response after palliative chemotherapy as the permeability surface value in patients within the responder group was significantly higher than within the non-re - sponder group. S670 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 5.8 SS 5.10 MRI of the gastric antrum for the quantification of Deep infiltrating endometriosis: can magnetic gastric motility: comparison between obese and normal resonance imaging predict the need for a colorectal weight patients surgeon? 1 1 2 3 3 S. Picchia , M. Rengo , M.A. Bali , D. Bellini , S. Badia , A. Brusic, S. Esler, L. Churilov, P. Chowdry, M. Sleeman, 1 1 2 3 A. Laghi ; Rome/IT, London/UK, Latina/IT P. Maher, N. Yang; Heidelberg, VIC/AU Purpose: To compare gastric motility in obese patients and normal weight Purpose: To identify Magnetic Resonance (MR) features that predict the need volunteers with magnetic resonance (MR). for colorectal (CR) surgeon intervention during gynaecologist-led laparoscopic Material and methods: This is a non-randomized single-centre study. The surgery for deep infiltrating endometriosis (DIE). motility analysis was performed on 1.5T MRI using 2DtrueFISP sequences on Material and methods: Retrospective cohort multivariate analysis. The cohort coronal and axial plane. Images were acquired before, immediately after a liq- consisted of women with laparoscopically-proven DIE, preoperative pelvic MR uid meal (Nutridrink plus 2 yolks, 1 albumen and 150 ml of water) and every 20 at Austin Health and subsequent surgery and histopathological confirmation at minutes for a total of 5 different acquisition times (T0-T5) and a total exam time Mercy Hospital (2006-2015, n=122). MRIs were reviewed independently by of 100 minutes. This 525-kcal meal was composed of 25% fat, 25% protein two radiologists blinded to outcomes. MR characteristics of pelvic endome- and 50% carbohydrate. Each gastric motility scan lasted 60 seconds, with 100 triosis were correlated to outcomes, to identify features associated with CR- images acquired in free breathing. Antral length (AL), maximal contraction am- surgeon intervention. plitude (MCA) and contraction frequency (CF) were evaluated on a dedicated Results: The readers demonstrated concordant findings for: presence of bow - software. Results obtained in obese patients and normal weight volunteers el lesions (kappa=0.68, 95%CI:0.55-0.81), bowel lesion length (Lin’s concord- were compared. ance coefficient 0.73, 95%CI:0.66-0.81); and bowel lesion depth (kappa=0.82, Results: Our final population consisted of twenty patients (10 obese and 10 95%CI:0.74-0.88). MR detected bowel lesions appeared more strongly associ- normal weight). MR showed that during fasting and postprandial period, in ated with the need for CR-surgeon intervention than current practice (defined obese patients all the evaluated parameters (AL, MCA and CF) were signifi - as the presence of a CR-surgeon in theatre), but this was not statistically sig- cantly lower than normal weight volunteers (mean basal AL obese nificant (area under receiver operating characteristic curve (ROC-AUC) 0.79 vs patients=6.75±0.8cm/normal weight patients=8.12±0.43cm, p value<0.05; 0.74, p=0.23). MR bowel lesions measuring ≥20mm or bowel lesions invading mean postprandial AL obese patients=7.56±1.03cm/normal weight the muscularis or submucosa, however, did demonstrate a statistically signifi - patients=9.4±0.12mm, p value<0.05; mean basal MCA obese cant correlation with CR-surgeon intervention and performed better than cur- patients=7.1±0.72mm/normal weight patients=8.97±0.43, p value<0.05; mean rent practice (ROC-AUC 0.84, p=0.014 and 0.90, p<0.0001, respectively). postprandial MCA obese patients=8.3±1.5mm/normal weight Conclusion: The current decision-making process to identify those patients patients=9.76±0.76mm, p value<0.05; mean basal CF obese with DIE that require a CR-surgeon to aid bowel resection is poorly defined and patients=8.7±0.98cpm/normal weight patients=10.3±0.9, p value<0.05; mean variable, resulting in sub-optimal resource allocation. This study identifies re - postprandial CF in obese patients=9.8±1.4cpm/in normal weight producible, objective MR features that better predict the need for CR-surgical patients=11.2±1.2cpm, p value<0.05). intervention than the current practice. This provides a basis for a prospective Conclusion: MRI identified differences in antral motility between obese pa - study for validation and development of a predictive scoring system to aid tients and normal weight volunteers. The obesity-related cause could become surgical planning and resource allocation. a potential new therapeutic target. SS 5.9 Utility of diffusion-weighted MRI with different b values in the evaluation of perianal fistulas M.H. Turkcanoglu, A.H. Yardımcı, C.T. Bektas, B. Kocak, O. Kilickesmez; Istanbul/TR Purpose: To determine the optimal b-value for diffusion-weighted imaging (DWI) in the diagnosis of perianal fistula and to demonstrate the diagnostic efficacy of this b-values and fistula tract thickness on DWI. Material and methods: Magnetic resonance (MR) images of 200 patients with the suspicion of perianal fistula referred to our Radiology Clinic were evaluated retrospectively. In our study, primary fistula tracts, internal orifices, abscess formations, horseshoe components and secondary tracts were evaluated. Axial fat-suppressed T2-weighted spin-echo (SE) images, DWI1 and DWI2 (b values, 400 and 1000s/mm ) contrast-enhanced T1-weighted images were analyzed for each patient by using a four-point scale. The combination of con- trast-enhanced T1-weighted images with T2-weighted images was used as reference standard. Results: 216 fistulas were detected in 200 patients (39 female,161 male). The sensitivity of DWI obtained at low b-value is higher in the evaluation of primary tract, internal orifice, secondary tract and horseshoe components in perianal fistulas. For the evaluation of abscess formations, it is seen that both low and high b-value DWI are equally sensitive. It has been shown that primary tract thickness is significant in detection of fistulas on DWI. The cut-off value for fistula thickness is 3.0 mm. Area under the curve 0.876 (0.831-0.922). Optimal cut-off value was obtained with 75.3% sensitivity, 100% positive predictive value, 100% specificity, 39.5% negative predictive value. Conclusion: The choice of optimal b values significantly affects the detection of perianal fistulas in DWI. Low b values (b=400s/mm(2)) had the best specific - ity and sensitivity. It has been shown that primary tract thickness is significant in detection of fistulas on DWI. S671 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 The Auditorium SS 6.3 Development and validation of a simplified magnetic Scientific Session SS 6 resonance index of activity for Crohn’s disease Small Bowel: imaging Crohn‘s disease J. Rimola, I. Ordas, S. Rodriguez, E. Ricart, J. Panés; Barcelona/ES SS 6.1 Purpose: To develop and validate a simplified and accurate magnetic reso - Diagnostic accuracy of layered pattern of enhancement nance index of activity (sMaRIA) for assessing activity and therapeutic re- during MRI exams in predicting active inflammation in sponse on patients with luminal Crohn’s disease (CD). Crohn’s disease: systematic review and meta-analysis Material and methods: Magnetic resonance enterography (MRE) data from 1 1 1 1 F. Rivosecchi , D. Bellini , D. Caruso , D. De Santis , 98 patients, including active and inactive segments (colon and terminal ileum), 1 2 1 2 M. Zerunian , A. Laghi ; Latina/IT, Rome/IT from two prospective studies were re-analysed to develop the sMaRIA using endoscopy (Crohn‘s disease endoscopic index of severity (CDEIS)) as the gold Purpose: To assess diagnostic accuracy of layered pattern (LP) on contrast- standard. Further analysis of responsiveness in a cohort of 37 patients who enhanced magnetic resonance imaging (MRI) in detecting active inflammation underwent MRE and endoscopy before and after 12 weeks of therapeutic in- in patients with Crohn’s disease. tervention was performed. Comparison between MaRIA and sMARIA for de- Material and methods: Methods for analysis were based on preferred report- tecting active/severe lesions and therapeutic response was performed. ing items for systematic reviews and meta-analyses (PRISMA). We searched Results: Logistic regression analysis showed that wall thickness >3mm, pres- MEDLINE, Cochrane library, Embase and Web of Science databases. Pooled ence of edema, ulcers and fat stranding were independent predictors of dis- estimates for sensitivity, specificity, pooled positive likelihood ratios (LR) and ease activity and were used therefore as descriptors of sMaRIA. The sensitiv- pooled negative LR were calculated using random effect model. I was used to ity and specificity of sMaRIA at segment level for detecting active disease evaluate heterogeneity. Quality assessment of diagnostic accuracy studies using a cutoff ≥5 were 90% and 81% (area under curve (AUC)=0.91, 95%CI (QUADAS) II was used to assess the quality of studies included. 0.88-0.94), and for detecting severe lesions (ulcers) using a cutoff ≥10 were Results: Of the 638 articles initially identified thorough databases searching, 7 85% and 92% (AUC=0.94, 95%CI 0.91-0.96). Correlation between sMaRIA were included for quantitative synthesis. Cumulative data for diagnostic ac- and CDEIS/MaRIA was excellent (r=0.84 and r=0.97, respectively; p<0.001). curacy on per patients basis were sensitivity 51% (95% CI: 44–59%; I : The sMaRIA accurately detected changes in lesion severity in response to a 78.5%), specificity 80% (95% CI: 74–85%; I : 91.4%), pooled positive LR 2.5 therapeutic intervention and was as reliable as endoscopy for the assessment 2 2 (1.32 to 4.73; I : 63.2%), pooled negative LR 0.67 (0.58 to 0.77; I : 0.0%). Het- of mucosal healing. erogeneity was high among studies. Conclusion: Simplified MaRIA index allows a faster and easier assessment of Conclusion: In conclusion, diagnostic accuracy of layered pattern on MRI has inflammation in CD by keeping high accuracy for both diagnosis and therapeu - a low sensitivity and a high specificity in detecting active inflammation in tic response. Main advantages over MaRIA includes, a less time consuming Crohn’s disease. Our results are limited due to the high heterogeneity among calculation and is not confounded by missing segments. studies included. SS 6.4 SS 6.2 Evaluation of image findings on CT enterography for Intra- and inter-observer variability in visually graded endoscopic complete remission after anti-tumor small bowel motility features from dynamic MRI of necrosis factor-alpha therapy in patients with Crohn’s Crohn’s disease patients 1 1 1 2 3 disease R. Gollifer , A. Menys , A. Plumb , F. Vos , J. Stoker , 1 1 1 2 S.H. Kim, J.E. Kim, T.O. Kim, J.-H. Yoon; Busan/KR D. Atkinson , S.A. Taylor ; London/UK, Delft/NL, Amsterdam/NL Purpose: To evaluate the image findings of CT enterography (CTE) obtained in patients with endoscopic complete remission (CR) after anti-tumor necrosis Purpose: Automated measurement of small bowel motility metrics is highly factor (TNF)-alpha therapy for Crohn’s disease. reproducible and is giving insights into the aetiology of abdominal symptoms Material and methods: Between August 2010 and October 2017, 36 con- in Crohn’s disease. It is unknown if a simple qualitative assessment of motility secutive patients with Crohn’s disease who received anti-TNF-alpha therapy by abdominal radiologists is sufficiently robust, and in particular is reproduci - and underwent pre-and post-therapy CTE as well as ileocolonoscopy were ble between observers. initially enrolled. CTE was performed with standard-dose enteric phase scan. Material and methods: 105 Crohn’s disease subjects underwent a 20-second Eleven patients were excluded due to a mismatch of follow-up intervals or a breath hold cine motility sequence, after mannitol ingestion, as part of mag- long interval (over 2 weeks) between CTE and ileocolonoscopy. Therefore, 25 netic resonance enterography. Five global small bowel motility metrics were patients (16 men, 9 women, mean age 26 years; range 18-43 years) were fi - visually graded by two experienced radiologists using a 10-point scale to as- nally analyzed. Two blinded readers reviewed pre-and post-therapy CTE im- sess 1) mean, 2) spatial variability, 3) temporal variability, 4) area of motile age findings in consensus as for active inflammation, i.e., mural hyperen - bowel and 5) distension quality. 26 datasets were analysed twice with a mini- hancement, mural thickening (thickness>3mm), mural stratifi cation, and mum gap of 2 weeks. Inter-and intra-reader variability was assessed using increased pericolonic fat attenuation in the rectum, colon, and terminal ileum, Bland-Altman plots. respectively. The endoscopic CR indicated mucosal healing identified by ile - Results: Inter-reader variability was generally poor for all motility features ocolonoscopy and served as the reference standard. graded e.g. the Bland-Altman mean difference in the spatial variability metric Results: Eleven patients had endoscopic CR. Six patients of them (6/11, 55%) was -0.76 au and the Bland-Altman 95% limits of agreement were -5.8 to +4.3 had residual mild mural thickening and mild mural hyperenhancement on post- across a range of values from 0.35 to 8.45. For radiologist 2, the worst agree- therapy CTE compared to pre-therapy CTE. The other 5 patients did not have ment was for the spatial variability metric with a mean difference between the any post-therapy CTE abnormalities. Non-endoscopic CR group (4 partial re- original grading and the repeat grading of -0.11 au and the Bland-Altman 95% missions and 10 poor remissions) showed a higher concordance rate of 86% adjusted limits of agreement were -4.1 to 3.9 across a range of values from 0.7 (12/14) with post-therapy CTE than endoscopic CR group did (45%, P=0.0358). to 7.4. Conclusion: Residual mild mural thickening and mild mural hyperenhance- Conclusion: Simple visual assessment of small bowel has poor inter and intra- ment were seen on post-therapy CTE over half of the patients having endo- observer agreement for most motility metrics and cannot replace automated scopic CR after anti-TNF-alpha therapy for Crohn’s disease. software measurement. S672 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 6.5 SS 6.7 Established damage detected at MRE in patients with Mural diffusion restriction pattern of small bowel: a Crohn’s disease in complete endoscopic remission possible tool for grading inflammation in Crohn’s J. Rimola, I. Alfaro, D. Vas, J. Castro, E. Ricart, J. Panés; disease Barcelona/ES N. Faluhelyi, O. Farkas, P. Bogner; Pécs/HU Purpose: To identify the Magnetic Resonance Enterography (MRE) lesions Purpose: To identify different diffusion restriction (DR) patterns in the small that persist in patients with Crohn’s disease (CD) in endoscopic remission as bowel wall and determine their diagnostic use in magnetic resonance enterog- indicators of established damage, and to determine its relationship with pre- raphy (MRE) examination of patients with Crohn‘s disease (CD). treatment MRE lesions. Material and methods: MRE (1.5 T GE Optima 450w MR) examinations were Material and methods: Patients with CD that had been included in prospec- performed after oral administration of mannitol (2.5%) solution. Diffusion- tive studies on autologous heamatopoyetic stem-cell transplant and/or anti- weighted images (axial DWI with b0, b400 and b800) and T2 images of 30 tumor necrosis factor (TNF) drugs were evaluated for this study. Inclusion cri- consecutive patients with small bowel CD activity were retrospectively ana- teria were: presence of at least one segment with severe inflammatory lesions lysed. In 8 patients, intravenous contrast agent (Gadovist) was also adminis- detected at MRE and achievement of endoscopic remission after 1 year of tered and pre- and postcontrast T1 images were added to the analysis. De- treatment. scriptive statistics, analysis of variance (ANOVA), Kruskal-Wallis and Results: 73 intestinal segments (28 patients) with severe inflammatory lesions Mann-Whitney tests were performed. at baseline achieved endoscopic remission after 1 year of treatment. The prev- Results: In 14 cases, DR was homogenously present in the inflamed bowel alence of creeping fat and bowel mural fat deposition did not change in asso- wall. Only mucosal DR was seen in 6 patients and DR showed a layered pat- ciation with the achievement of endoscopic remission (p=0.34 and p=0.35 re- tern (both mucosal and serosal restriction with a central band of relatively spectively). Furthermore, luminal strictures persisted in 50% of segments with lower restriction) in 10 cases. Restriction patterns were corresponding to the initial strictures, and wall thickness remained >3mm in 29% of segments with contrast enhancement pattern of all additional contrast-enhanced cases. Pa- severe inflammatory lesions at baseline. Regression analysis showed that pre - tients with layered pattern had a significantly higher Clermont score (median: dictive lesions on pre-treatment MRE for established residual mural lesions 31.68, range: 3.87) than the rest of the patients (median: 21.12 range: 18.94, after achieving endoscopic remission were the presence of bowel fat deposi- p=0. 04746). The three groups proved to have a different mural thickness with tion (OR=48.3, p=0.001) and strictures at baseline (OR=15.96, p=0.004), the thickest wall in the “layered” group (mean: 8.7 mm, SD: 0.82, p<0.00001). whereas creeping fat was the only predictive factor for persistent extramural Conclusion: The well-known contrast enhancement patterns of bowels in ac- lesions (OR=35.8, p<0.001) and for either mural or extramural lesions tive CD (homogenous, mucosal and layered) are also identifiable as DR pat - (OR=16.25, p<0.001). terns on DWI-MRE. Layered restriction pattern can be a marker of more severe Conclusion: Up to 41% of bowel segments with initial severe inflammation in activity of small bowel inflammation in Crohn‘s patients. long-standing endoscopic remission after treatment had residual lesions at MRE. Presence of bowel fat deposition, strictures and creeping fat at pre- SS 6.8 treatment MRE may predict the development of established damage. The utility of diffusion-weighted imaging in the characterisation and assessment of activity of small SS 6.6 bowel and colonic Crohn’s disease on MR enterography Preoperative evaluation of small bowel complications in J.P. Walsh, D. Byrne, H. Fenlon, C. Cronin; Dublin/IE Crohn‘s disease: comparison of diffusion-weighted and Purpose: To demonstrate the utility of diffusion-weighted imaging (DWI) in contrast-enhanced MR imaging characterisation and assessment of activity of small bowel and colonic Crohn’s 1 2 1 1 3 M. Barat , C. Hoeffel , M. Bouquot , R. Dautry , M. Boudiaf , disease on MR enterography (MRE). 1 1 1 1 1 K. Pautrat , R. Kaci , C. Eveno , M. Pocard , P. Soyer , Material and methods: Of 145 patients who underwent MRE over an 18 1 1 2 3 A. Dohan ; Paris/FR, Reims/FR, Meudon la Foret/FR month period (July 2016 – December 2017), we retrospectively analysed imag- Purpose: To compare diffusion weighted-imaging (DW) magnetic resonance ing performed for 63 consecutive patients (39 female, 24 male) with biopsy- imaging (MRI) to contrast-enhanced (CE) MRI in the preoperative assessment proven Crohn’s disease. Appearances on T2 weighted, T1 contrast-enhanced of small bowel complications of Crohn‘s disease (CD) with MR enterography sequences and DWI, where available, were assessed. Using a semi-quantita- (MRE). tive technique, MRE activity scores were calculated for involved segments. Material and methods: Thirty-eight patients who underwent surgery for CD Results were compared with endoscopy and histologic findings, where avail - complications and preoperative MRE were included. MRE examinations were able. blindly analyzed independently by one junior and one senior abdominal radi- Results: Of 71 segments analysed, 58 demonstrated abnormality on contrast- ologist for the presence of fistula, stenosis and abscesses. During a first read - enhanced or DWI imaging (50 small bowel, 8 colon). Abnormal contrast en- ing session T2-weighted images (T2-WI), steady-state sequences (True-FISP) hancement was identified in 48 of 50 abnormal small bowel segments and 8 of and DW-MRI were reviewed (set-1) and during a separate distant session, T2- 8 colonic segments. Abnormal DWI signal was identified in 25 of 31 small WI, True-FISP and CE-MRI were reviewed (set-2). Performances of each read- bowel segments and 8 of 8 colonic segments. DWI was unavailable for 26 seg- er for the two sets were evaluated using imaging, intraoperative and patho- ments. Calculated sensitivity of contrast-enhanced MRE for active small bow- logical findings as the standard of reference. el inflammation was 96.0% [86.3-99.5%], while that of DWI was 80.7% [62.5- Results: Forty-eight fistulas, 43 stenosis and 11 abscesses were deemed pre - 92.6%]. Analysis of colonic activity was confounded by small sample size. sent using the standard of reference. For the senior radiologist, sensitivities for Conclusion: While there was a trend towards greater sensitivity for contrast- the detection of fistulas, stenosis and abscesses ranged from 80% to 100% enhanced MRE over DWI for detection of acute small bowel inflammation, this for set 1 and from 88% to 100% for set 2 and specificity ranged from 56% to did not reach the level of statistical significance. While an ideal MRE protocol 70% for set 1 and from 53% to 93% for set 2, with no significant differences would likely include both sequences, this study suggests that DWI may pro- between the different sets (P-values, 0.342-0.429). For the junior radiologist, vide an alternative to contrast-enhanced sequences, especially for longitudinal higher performances were obtained with set 2 (P-values, 0.001 - 0.007). assessment of disease activity. Conclusion: For a senior radiologist, DW-MRE has similar sensitivities than CE-MRE for the diagnostic of CD complications. For a junior radiologist, CE- MRE yields better results than DW-MRE. S673 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 6.9 Magnetic resonance enterography (MRE) and enteric ultrasound (US) in newly diagnosed or relapsing Crohn’s disease: patient acceptability, perceived burden and preferences S.A. Taylor, A. Miles, T. Metric Investigators; London/UK Purpose: Magnetic resonance enterography (MRE) and ultrasound (US) are widely used to image Crohn’s disease. The aim was to compare patient ac- ceptability and burden of MRE and US to each other, and to other enteric in- vestigations, and identify drivers of preferences. Material and methods: 159 patients (mean age 38, 94 female) prospectively recruited to a multicentre diagnostic accuracy study comparing MRE and US completed a detailed experience questionnaire pertaining to the burden and acceptability of small bowel investigations, including MRE and US. Data was compared t-tests, chi-square tests and McNemar tests as appropriate. Predic- tors of burden and patient preference were tested using linear and logistic re- gression. Results: US, MRE and colonoscopy were rated as very/fairly acceptable by 98.6%, 88.3% and 60.0% of patients respectively (p<0.001). MRE recovery time was significantly longer than US (p<0.001), but shorter than colonoscopy (p<0.001). Patients were less willing to undergo MRE again than US (90.7% vs. 98.5%, p=0.012, but more willing than for colonoscopy (74.7%, p=0.017). MRE resulted in greater burden than US (t=13.56, df=144, p<0.001), although overall burden scores were low. Younger age and significant levels of emo - tional distress were associated with MRE burden. A majority (80.0%) preferred US to MRE. Higher MRE discomfort predicted a preference for US but patients rated test accuracy as more important than scan discomfort. Conclusion: MRE and US are well tolerated, although MRE generates greater patient burden, longer recovery times, and is less preferred than US. Patients, however, place greater emphasis on diagnostic accuracy than scan burden when determining preferences. SS 6.10 Accuracy of emergency CT examination without bowel preparation in the assessment of disease activity and complications in ileocolonic Crohn’s disease patients compared to standard magnetic resonance enterography S.G. Drago, S. Lombardi, A. Casiraghi, C. Talei Franzesi, S. Sironi, D. Ippolito; Monza/IT Purpose: To compare the agreement between abdomino-pelvic CT performed with intravenous contrast material without oral contrast administration, in an urgent setting at the Emergency Department, in comparison to MR-enterogra- phy (MRE) in detecting inflammatory bowel disease findings and its complica - tions in Crohn’s disease (CD) patients. Material and methods: Forty-eight patients with known or suspicion of CD underwent abdomino-pelvic contrast-enhanced CT study on a 256-row MDCT (iCT, Philips Medical Systems), without any previous bowel distention. After- ward, the same group of patients underwent MR-Enterography on a 1.5 T (Philips, Achieva MR system) using oral administration of 1.5Lt of contrast agent and intravenous contrast. Two readers independently reviewed all the images of both techniques and analyzed the following cross-sectional find - ings: bowel wall thickness and enhancement, mesenteric lymph-nodes, vas- cular engorgement, fibro-fatty proliferation, fistulas and abscesses. Results: Two readers assessed 56 diseased bowel segments with both mo- dalities: small bowel, colon-sigmoid and rectum. Inter-observer agreement was higher for bowel obstruction (k=0.650 CT, k=0.671 MRE), its complica- tions (k=0.810 CT, k=1 MRE) and lymphoadenopathy (k=0.810 CT, k=0.690 MRE). Higher inter-reader agreement was achieved for CT in comparison with MRE for wall thickening (k=0.750 CT, k=0.610 MRE), fibro-fatty proliferation (k=0.780 CT, k=0.523 MRE) and vascular comb sign (k=0.610 CT, k=0.360 MRE). Conclusion: Abdominal contrast-enhanced CT without bowel distention of- fers high diagnostic value in the assessment of bowel wall involvement and its complications in CD and, in selected cases, immediate MR-Enterography ex- amination could be postponed and reserved for follow-up. S674 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 The Liffey A SS 7.3 Supersonic B-ratio mode and shear-wave elastography Scientific Session SS 7 performances in detection and graduation of steatosis Diffuse and chronic liver diseases: current and liver fibrosis in patients with liver transplantation M. Dubois, V. Brun, M. Rayar, M. Auger, L. Beuzit, assessment of liver fibrosis and steatosis E. Quehen, Y. Gandon, B. Turlin, P. Houssel-Debry, A. Paisant; Rennes/FR SS 7.1 Purpose: To evaluate the diagnostic performance of Supersonic B-ratio mode Doppler ultrasonography devices, including and shear-wave elastography (SWE) for the assessment of steatosis and liver elastography, allows accurate screening for severe fibrosis in patients with liver transplantation. fibrosis Material and methods: After institutional review board approval and written P. Bazeries, J. Delahaye, V. Cartier, J. Lebigot, J. Boursier, informed consent, patients hospitalized from June 2017 to December 2017 for C. Aubé; Angers/FR a systematic check-up after liver transplantation were prospectively included. Purpose: To evaluate the contribution of the elastography following a Mode B Each patient underwent the same day SWE, liver biopsy (LB) and biological ultrasonography examination in the screening of severe hepatic fibrosis exams. Steatosis was measured using B-Ratio mode between the liver and (METAVIR ≥ F3). kidney and liver stiffness with 3 SWE consecutive measures. LB was used as Material and methods: Five hundred and fourteen patients were prospec- gold standard, grading steatosis S0 (0%), S1 (1-10%), S2 (11-30%) or S3 tively included in a monocentric six years study. All patients underwent a liver (>30%), and grading liver fibrosis from F0 to F4 according to the Metavir score. biopsy to establish the fibrosis stage (METAVIR scoring). The signs of hepatic Results: Fifty patients were included. Mean B-ratio value was significantly dysmorphism and portal hypertension were studied in ultrasonography ex- higher in patients with steatosis (0.94±0.13 vs 1.34±0.44, p<0.001). Sensitivity amination with Doppler. Liver stiffness evaluation was performed using both and specificity were, respectively, 77.3% (0.56-0.90) and 80% (0.58-0.92) the acoustic radiation force impulse (ARFI) (Siemens) and 2D shear-wave elas- (area under the receiver operating characteristic (AUROC)=0.82) for S≥1 with a trography (SWE - Supersonic Imagine) techniques. B-ratio cut-off value of 1.02; 84.6% (0.56-0.97) and 85.7% (0.68-0.95) (AU- Results: Three ultrasound parameters were independent predictors of severe ROC=0.92) for S≥2 with a B-ratio cut-off value of 1.13; and 85.7% (0.46-0.99) fibrosis: liver surface irregularity, spleen length (≥110 mm) and demodulation of and 76.5% (0.60-0.88) (AUROC=0.80) for S≥3 with a B-ratio cut-off value of the hepatic veins waveform. The presence of at least one of these 3 parame- 1.20. Mean SWE value for patients without significant fibrosis (≤F1) was ters had 85.6% sensitivity and 36.1% specificity for the diagnosis of severe 14.1±9.4 vs 27.4±0.31 for patient with fibrosis. fibrosis. The use of ARFI (diagnostic cut off ≥1.59m/s) after the detection of Conclusion: Supersonic B-ratio mode is a non-invasive and accurate method one of this three signs dramatically increased the specificity at 80.8%, at a to detect and graduate steatosis in patients with liver transplantation. Fibrosis price of a slight decrease in sensitivity (73.7%). Same results were obtained cut-off values seem to be higher in transplanted liver than in normal liver. with SWE (diagnostic cut-off ≥9.5kPa) with 73.3% specificity and 81.5% sen - sitivity. Addition of an elastographic measurement allows obtaining a positive SS 7.4 predictive value of approximately 50% for the diagnosis of severe fibrosis, A comparative evaluation of modified DIXON, CT (LAI) which is an acceptable rate to refer patients to a hepatologist for a complete and US elastography, in quantification of liver fat, with assessment of chronic liver disease. MR spectroscopy as reference Conclusion: The screening of severe hepatic fibrosis using 3 simple ultra - R. Jain, S. Kale, N. Panchal, N. Kundaragi, sound signs with addition of elastographic measurement in case of positivity B. M, R. Talwade, N. Reddy; Bangalore/IN of one of them is a feasible and accurate procedure. Purpose: Early diagnosis and treatment of fatty liver disease, along with mod- ifications to lifestyle, helps reduce the rate of progression to end stage liver SS 7.2 disease and hepatocellular cancer. Additionally, liver fat is a risk for post-oper- Diagnostic performance of texture analysis for the ative complications of a transplant. Non-invasive MRI can be used to monitor assessment of hepatic fibrosis in patients with liver fat. This study was conducted to compare the accuracy of modified DIX - nonalcoholic fatty liver disease (NAFLD): preliminary ON, CT(liver attenuation index (LAI)) and US Elastography, with MR Spectros- results and comparison with MR elastography (MRE) 1 2 2 2 2 copy (MRS) as a reference benchmark in quantification of liver fat. R. Cannella , M. Tublin , A. Borhani , J. Behari , A. Furlan ; 1 2 Material and methods: Tests were performed on 80 patients (35 females and Palermo/IT, Pittsburgh, PA/US 45 males) which were grouped into hepatosteatosis-alcoholic (n=15), non-al- Purpose: To investigate the performance of texture analysis (TA) in the quanti- coholic non-obese (n=8), obese (n=23), liver donors (n=25) and post-chemo- fication of fibrosis in NAFLD and to compare it with MR elastography (MRE). therapy (n=9). MRI images were obtained on 3T Philips Ingenia, CT indices Material and methods: 54 adult patients (33 females, 21 males) with biopsy- calculated on Philips 128 Slice Ingenuity Core Scanner, US Elastography per- proven NAFLD were prospectively enrolled and underwent MRI on a 1.5T sys- formed on Philips Affinity70. Results were independently interpreted by two tem using a combination of T1-weighted gradient-recalled-echo (T1W 3D- radiologists. All data and interpretations were compiled and compared for pa- GRE) sequence (slice thickness =4.6 mm) and 2D-MRE. Texture analysis was tient groups, CT- LAI, elastography score (F0-F4), percentage of liver fat performed using a commercially available research software (TexRAD) on axial through mDIXON and MRS. T1W images. A circular region of interest (ROI) (3 cm ) was traced on an image Results: Sensitivity values are 97% for MR Spectroscopy, 96.5% for mDIXON, at the level of the porta hepatis. Receiver operating curves (ROC), areas under 71% for CT (LAI) and 64% for US elastography. The study showed excellent the ROC (AUROC) and 95% confidence intervals were calculated to assess correlation between MR spectroscopy and mDIXON (r value of 0.93), moder- the accuracy of each textural parameter and of liver stiffness for the diagnosis ate correlation with CT-LAI (r value of 0.89) and poor correlation with US elas- of significant fibrosis (SF: F≥2) and advanced fibrosis (AF: F≥3). AUROCs were tograpy (r value of 0.003). 15 out of 80 patients underwent biopsy, the histo- compared using the DeLong test. pathology results for these were included in the study. Results: Thirty-seven subjects had SF and 20 had AF. The textural parameters Conclusion: Modified DIXON is as accurate as MR spectroscopy as a nonin - with the best performance were SD and entropy with AUROC 0.755 (0.619- vasive method for quantification of hepatosteatosis, compared to the widely 0.862, p≤0.0002) and 0.769 (0.634-0.873, p<0.0001) for SF and AUROC 0.746 used CT (LAI) and US elastography which have their inherent drawbacks of (0.609-0.854, p≤0.0004) and 0.754 (0.618-0.861, p≤0.0002) for AF. The AU- radiation and operator dependence. ROC of MRE for SF was 0.785 (0.652–0.885, p≤0.0001) and for AF was 0.923 (0.817-0.978, p≤0.0001). While the AUROC of MRE for AF was significantly higher than that of SD and entropy (p=0.03 and p=0.04), there was no signifi - cant difference among AUROCs for SF (p=0.85 and p=0.74). Conclusion: Texture analysis has fair accuracy for the quantification of he - patic fibrosis in NAFLD, although the performance is inferior to MRE for the assessment of AF. S675 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 7.5 SS 7.7 Clinical value of ultrasonic adaptive sound speed Reliability criteria for liver stiffness measurement with estimation for the diagnosis and quantification of acoustic radiation force impulse 1 2 1 1 2 hepatic steatosis J. Boursier , C. Cassinotto , V. Cartier , A. Lannes , S. Shili , 1 2 1 1 1 2 1 2 M. Dioguardi Burgio , M. Imbault , M. Ronot , A. Faccinetto , J. Lebigot , B. Lapuyade , P. Cales , J.-B. Hiriart , 1 1 1 2 1 3 2 1 B. Van Beers , P.-E. Rautou , L. Castera , J.-L. Gennisson , S. Michalak , B. Le Bail , A. Mouries , F. Oberti , 2 1 1 2 1 2 2 1 M. Tanter , V. Vilgrain ; Clichy/FR, Paris/FR I. Fouchard , F. Chermak , V. De Ledinghen , C. Aubé ; 1 2 3 Angers/FR, Pessac/FR, Bordeaux/FR Purpose: To evaluate the ability of a new ultrasound (US) method based on Sound Speed Estimation (SSE) in detection, quantification, and grading of he - Purpose: To determine the reliability criteria of liver stiffness measurement patic steatosis using magnetic resonance (MR) proton density fat fraction (LSM) using acoustic radiation force impulse (ARFI) in chronic liver diseases. (PDFF) as a reference standard and to calculate one US fat index based on the Material and methods: 1094 patients with chronic liver diseases had liver bi- patient’s SSE. opsy and LSM with ARFI in two centers. Advanced fibrosis was defined as Material and methods: We consecutively included N=50 patients as study non-alcoholic steatohepatitis clinical research network (NASH CRN) histologi- cohort and further N=50 as validation cohort who underwent both SSE and cal scoring system F≥3 or Metavir F≥2. abdominal MR. Hepatic steatosis was classified according to MR-PDFF cut - Results: LSM failure rate was 0.9% (10/1094). Among the remaining 1084 offs as S0 ≤6.5%, S1 from 6.5 to 16.5%, S2 from 16.5 to 22% and S3≥ 22%. patients (male: 61.9%; age: 54.3±13.3 years), 48.5% had advanced fibrosis Receiver operating curve analysis was performed to evaluate the diagnostic and 16.8% cirrhosis. Etiologies of chronic liver disease were non-alcoholic performance of SSE in diagnosis of steatosis. Based on the optimal data fit fatty liver disease (NAFLD) (48.5%), viral hepatitis (26.3%), alcohol (12.2%), derived from our study, we proposed a correspondence between the MR- and other (13.0%). ARFI accuracy decreased with increasing interquartile PDFF and an US fat index. Coefficient of determination R was used to evalu- range (IQR) of liver stiffness to the median (IQR/M) ratio, especially for interme- ate fit quality and was considered as robust when R >0.6. diate/high levels of liver stiffness. Three reliability categories were thus de- Results: Study and validation cohort presented mean SSE value of fined: “very reliable” (IQR/M <0.15), “reliable” (0.15≤ IQR/M <0.35 or IQR/M 1.570±0.026 and 1.568±0.023 mm/µs for S0 and 1.521±0.031 and 1.514±0.019 ≥0.35 with ARFI median <1.37 m/s), and “poorly reliable” (IQR/M ≥0.35 with mm/µs for S1-S3 (p<0.01) patients, respectively. SSE threshold of ≤1.537 mm/ ARFI median ≥1.37 m/s). Using these criteria, the rates of patients correctly µs had 80% sensitivity and 85.7% specificity in diagnosis of steatosis in the classified were, respectively: 80.9%, 73.7%, and 57.8% for advanced fibrosis study cohort. Robust correspondence between MR-PDFF and the US fat in- (p=0.029 between very reliable and reliable; p <0.001 for other paired com- dex was found both for the study cohort (R = 0.73) and the validation cohort parisons); and 92.6%, 83.4%, and 50.0% for cirrhosis (p<0.001 for all com- (R = 0.76). parisons). 23.6% of the ARFI examinations were very reliable, 55.0% reliable, Conclusion: SSE can be used to detect, quantify and grade liver steatosis and and 21.4% poorly reliable. The skin-liver capsula distance was an independent to calculate an US fat index. predictor of poorly reliable LSM, which occurred in 52.7% of patients having a distance ≥30mm. Conclusion: Based on IQR/M ratio and ARFI median, we defined three reliabil - SS 7.6 ity categories for LSM using ARFI, associated with different diagnostic accura- Transducer-free hepatic magnetic resonance cies. These new reliability criteria will increase confidence in ARFI results and elastography using cardiac wave induction at 0.3 ms diffusion of this technique. temporal resolution 1 2 1 1 M.A. Troelstra , J.H. Runge , A. Polcaro , O. Darwish , 1 3 2 1 J. De Arcos , T. Schneider , A.J. Nederveen , R. Sinkus ; SS 7.8 1 2 3 London/UK, Amsterdam/NL, Guildford/UK Non-invasive monitoring of hepatic steatosis via acoustic structure quantification of US with MR Purpose: This study aims to develop a transducer-free hepatic magnetic reso- spectroscopy as a reference standard nance elastography (MRE) method to facilitate clinical MRE implementation. D.H. Lee, J.Y. Lee; Seoul/KR We propose a cardiac-triggered motion-sensitised pencil-beam, for capturing cardiac-induced hepatic shear-waves at a high temporal resolution (>0.3 ms). Purpose: To prospectively evaluate whether monitoring hepatic steatosis by Material and methods: Six healthy volunteers were scanned on a Philips 3T- ultrasonography with an acoustic structure quantification (ASQ) technique is scanner. The pencil-beam navigator was positioned below the heart in liver feasible when using magnetic resonance spectroscopy (MRS) as a reference segments 2-4. ECG-triggered sequences obtained 35-40 pencil-beam images standard. (d=30mm, L=80mm, t=20ms) acquired in one R-R interval and repeated for 60 Material and methods: Thirty-six patients with suspected fatty liver disease successive R-R intervals, divided over four breath-holds, attaining temporal underwent both ultrasonography with ASQ and MRS on the same day. After a resolutions >0.3 ms. Phase-images yielded space-time images of hepatic mean follow-up period of 11.4±2.5 months, follow-up ultrasonography with transient shear-wave propagation. Straight lines were fitted along phase per - ASQ and MRS were performed on 27 patients to evaluate whether hepatic st nd turbations at 1 and 2 heart sound, to determine shear wave speed and shear steatosis improved. The focal disturbance (FD) ratio, as calculated using ASQ, stiffness μ=ρc (assuming no loss-effects). and the hepatic fat fraction (HFF), estimated by MRS, were obtained at both Results: Analysis showed two shear-wave propagations from the heart into initial and follow-up examinations. Pearson’s correlation coefficient was calcu - st nd the liver, at 1 (atrioventricular-valve closure) and 2 heart sounds (aortic-/ lated to assess correlations between ordinal values. nd pulmonary-valve closure). 2 heart sound waves showed an average wave Results: The FD ratio showed a strong, negative linear correlation with the speed of 1.49+/-0.07 m/s and stiffness of 2.24+/-0.21 kPa, 2x higher shear HFF after logarithmic transformation of both variables from the initial examina- nd stiffness compared to classical-MRE at 40Hz. The 2 heart sound shows a tions of 36 patients (ρ=-0.888; P<0.001) and the follow-up examinations of 27 frequency emission spectra peak at 85Hz. Liver-tissue properties are approxi- patients (ρ=-0.920; P<0.001). There was also a significant, negative linear cor - mately linear with frequency, resulting in stiffnesses of ~1kPa measured at relation between the change in the logarithm of the FD ratio and the change in 40Hz increasing to ~2kPa at 85Hz. the logarithm of the HFF by MRS over the follow-up period (ρ=-0.645; Conclusion: This approach allows transducer-free visualisation of hepatic P<0.001). In 16 patients with increased FD ratio on follow-up, HFF on follow- cardiac-induced shear-waves at high temporal resolution (>0.3ms), using up MRS significantly decreased, and HDL level significantly increased whereas waves created by cardiac-valve closure. Results show fast wave propagation, LDL tended to decrease. caused by high-frequency content of waves generated from cardiac-valve clo- Conclusion: The FD ratio was significantly correlated with the HFF at both the sure. This transducer-free technique could facilitate clinical MRE implementa- initial and follow-up examinations, and there was also a significant correlation tion, once it is validated in a larger population. between the changes in the FD ratio and the changes in the HFF over the fol- low-up period. S676 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 Liffey Hall 2 SS 7.9 Molecular features, histological parameters and Scientific Session SS 8 enhancement measurements in nonalcoholic steatohepatitis patients using gadoxetic acid-enhanced Malignant pancreatic tumours: new trends MR Imaging (preliminary results) 1 2 1 1 N. Bastati-Huber , D.S. Feier , A. Beer , S. Pötter-Lang , SS 8.1 1 1 1 H. Einspieler , A. Ba-Ssalamah ; Vienna/AT, Lower energy levels and iodine density images increase Cluj-Napoca/RO pancreatic tumor conspicuity on rapid kV-switching dual Purpose: To assess the relationship of organic anion transporting polypeptide energy CT (OATP1B1/3) expression with histological parameters and relative enhance- S. Aslan, M.S. Nural, İ. Camlidag; Samsun/TR ment ratio in patients with nonalcoholic fatty liver disease (NAFLD), defined Purpose: To evaluate the efficacy of rapid kV-switching dual energy CT (rs - according to the steatosis activity and fibrosis (SAF) scoring system, which is DECT) energy level optimization and iodine density (ID) images on pancreatic based on the semiquantitative scoring of steatosis activity and liver fibrosis. cancer detection and conspicuity. Material and methods: The local institutional review committee approved this Material and methods: 56 patients with pancreatic adenocancer were pro- study and waived written informed consent. This was a retrospective study of spectively enrolled and underwent rsDECT in pancreatic protocol. Largest le- gadoxetic acid-enhanced 3T MR imaging performed in 52 consecutive pa- sion diameters (LLDs) were measured on 45 keV, optimal CNR, 70 keV and ID tients with NAFLD (mean age (SD), 50.46 (16.52) years. The MR images were images. Attenuation differences between the lesion and normal parenchyma analyzed by using the relative enhancement (RLE) (the ratio of signal intensities (HU), image noise for each energy level, contrast gain and lesion contrast-to- of the liver parenchyma before and 20 minutes after intravenous administration noise ratio (CNR) were calculated. Iodine content of the lesions and the paren- of gadoxetic acid). Univariate and multiple regression analyses were applied to chyma were measured on ID images. Isodense lesions were treated as a sub- identify variables associated with OATP1B1/3 expression. group and all measurements were also applied to them. All analyses were Results: OATP1B1/3 expression correlated with RLE (r=0.49, p=0.0002), the performed by two radiologists at different times and inter-reader agreement degree of liver fibrosis (r=-0.27, p=0.04) but not with steatosis (r=-0.16, was evaluated. p=0.23), ballooning (r=-0.22, p=0.1) or inflammation (r=-0.07, p=0.57). Accord - Results: Optimal CNR was 52 ± 8,5 kEv. LLDs were measured on ID images ing to multivariate analysis both RLE and liver fibrosis are independent predic - (p< 0.001). Attenuation measurements on 45 keV and optimal keV were sig- tors of OATP1B1/3 expression (coefficient of determination R2=0.32, p=0.01). nificantly higher than 70 keV images. Attenuation difference between the lesion Conclusion: The degree of OATP1B1/3 expression correlated statistically with and parenchyma was significantly different between 70 keV images and other gadoxetic acid relative enhancement and the degree of liver fibrosis in patients energy levels but not different between 45 keV and optimal CNR. Highest im- with NASH. age noise was calculated in 45 keV. Highest contrast gain was between 70 keV and optimal CNR and highest CNR was measured in optimal CNR. Iodine con- SS 7.10 tent of tumors were significantly lower than the parenchyma (1,83 mg/cc vs The positive effects of bariatric surgery on non-alcoholic 3,05 mg/cc, p < 0.05). Isodense tumors also showed similar results. Inter-read- fatty liver disease: evaluating the changes in the liver fat er agreement was high (ICC 0,78 - 0,93). fraction and liver dimensions by using ideal IQ Conclusion: rsDECT is a reliable method for pancreatic cancer detection and sequences delineation, particularly isodense tumors. Y. Metin, N. Orhan Metin, O. Özdemir, M.E. Kadıoğlu, E. Beykoz Çetin, S. Kalcan, M.K. Çolakoğlu; Rize/TR SS 8.2 Purpose: To evaluate the early and intermediate effects of bariatric surgery on Prediction of residual tumor classification and overall liver fat fraction, liver length and liver volume using ideal IQ sequences at 3T survival in pancreatic cancer by preoperative CT 1 1 1 2 1 MRI. J.S. Bae , J.H. Kim , I. Joo , W. Chang , J.K. Han ; 1 2 Material and methods: Body mass index (BMI), body weight, liver fat fraction Seoul/KR, Seongnam/KR (L-FF), liver volume and craniocaudal length were measured preoperatively (1 Purpose: To predict resection (R) stage classification and overall survival on week before surgery) and postoperatively (1 and 6 months after surgery) in 46 preoperative CT in patients who underwent surgery for pancreatic cancer. patients (31 female, 15 male; age range, 21–60 years) who underwent bariatric Material and methods: In this retrospective study, 216 patients with pancre- surgery between May 2016 and August 2017. Liver volumes, L-FF were calcu- atic cancer who underwent CT and surgery were included. Preoperative CT lated from Ideal IQ sequences. findings were assessed by two radiologists and R classification was catego - Results: BMI decreased from 46.3 ± 5.5 to 40.7 ± 5.3 (at first month), and to rized into no residual tumor (R0) and residual tumor (R1 or R2). We assessed 32.6 ± 5.5 kg/m at the end of sixth month. Mean liver volume decreased from the correlation between CT findings and R classification. In addition, we used 2641.2 ± 543.6 to 1954.1 ± 384.2 cm (at first month) and to 1776.3 ± 361.1 Kaplan-Meier estimation for survival analysis and Cox proportional hazard cm at the end of sixth month. L-FF decreased from 16.4 ± 10.2 to 7.2 ± 4.9% model to find prognostic factors for overall survival. (at first month) and to 4.0 ±2.7% at the end of sixth month. At the first month, Results: There were 153 patients in R0 group and 63 patients in R1 or R2 60.9% of patients and at the sixth month 82.6% of patients had resolution of group. Tumor size (odds ratio (OR) 1.045, 95% confidence interval [CI]: 1.005- steatosis. 1.086), definite peritumoral fat stranding (OR 3.826, 95% CI: 1.765-8.293), Conclusion: A significant decrease in liver volume, liver length and FF was portal vein involvement (OR 2.795, 95% CI: 1.216-6.423), and suspicious dis- achieved at first and sixth months after bariatric surgery. The reduction in L-FF tant metastasis (OR 2.916, 95% CI: 1.296-6.561) were independent predictors is most prominent after the first month and continues at sixth month. The pos - for residual tumor (P<0.05). On survival analysis, median survival period was itive effects of bariatric surgery on liver dimensions and FF can be monitorized 29.2 months in R0 group and 12.2 months in R1 or R2 group. T4 stage (hazard by using ideal IQ sequences. ratio (HR) 4.330, 95% CI: 1.208-2.555), N1 stage (HR 1.757, 95% CI: 1.208- 2.555), common hepatic artery involvement (HR 4.821, 95% CI: 2.125-10.934) and R stage (HR 2.499, 95% CI: 1.649-3.789) were predictors of poor survival (P<0.05). Conclusion: Preoperative CT is useful to predict R classification using tumor size, definite peritumoral fat stranding, portal vein involvement, and suspicious distant metastasis, as well as to anticipate poor survival using T stage, N stage, common hepatic artery involvement and R classification. S677 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 8.3 SS 8.5 Radiological assessment of local resectability status in Perfusion measurements in pancreatic cancer by means patients with pancreatic cancer: interreader agreement of intravoxel incoherent motion MRI and CT perfusion: and reader performance at two different classification correlation with each other and with histological systems microvessel density 1 1 1 2 N. Kartalis , C. Valls , E. Axelsson , M. Andersson , P.F.J. Mayer, F. Fritz, S. Skornitzke, W. Stiller, M. Klauss, 3 4 5 1 I. Keussen , J. Strinnholm , W. Bartholomä , M. Del Chiaro , H.U. Kauczor; Heidelberg/DE 1 1 1 1 R. Segersvärd , L. Lundell , L. Loizou ; Stockholm/SE, Purpose: The aim of this study was to compare intravoxel incoherent motion 2 3 4 5 Gothenburg/SE, Lund/SE, Umeå/SE, Linköping/SE (IVIM) diffusion-weighted (DW) MRI and CT perfusion to assess tumor perfu- Purpose: To assess the interreader agreement and reader performance in the sion of pancreatic ductal adenocarcinoma (PDAC). evaluation of patients with pancreatic cancer (PC) at two classification sys - Material and methods: In 19 patients with PDAC, DW MRI and CT perfusion tems of local resectability status, namely the national comprehensive cancer were conducted on the day before surgery. IVIM analysis of DW MRI was per- network (NCCN) and Karolinska Institutet’s classification system (KICS). formed and the parameters perfusion fraction f, pseudodiffusion coefficient D*, Material and methods: In this institutional review board-approved retrospec- and diffusion coefficient D were extracted. The CT perfusion parameters blood tive study, six radiologists evaluated independently pancreatic CT-examina- volume (BV) and blood flow (BF) were estimated with a deconvolution-based tions of 30 patients randomly selected from a tertiary referral centre’s multidis- analysis. In 10 patients, intratumoral microvessel density (MVD) and microves- ciplinary tumour board database. Based on well-defined criteria of sel area (MVA) were analyzed from representative immunostained tissue slides, tumour-vessel relationship, each patient was assigned into one of three NCCN using an anti-CD34 antibody as endothelial marker. Pearson correlation coef- and six KICS categories. We assessed the intraclass correlation coefficient ficients between these parameters were calculated. (ICC; a score of 0–0.2 indicates poor agreement, 0.3–0.4 fair agreement, 0.5– Results: There were significant positive correlations between MVD/ MVA and 0.6 moderate agreement, 0.7–0.8 strong agreement, and >0.8 a very strong f/ BF (r ≥ 0.666, p ≤ 0.036). f significantly positively correlated with BF and BV agreement) and compared the percentages of correct tumour classification of (r ≥ 0.691, p ≤ 0.001). Correlation coefficients between BF and MVD/MVA were the six readers at both systems (Chi-square test; a P-value <0.05 was consid- not significantly different from correlation coefficients between f and MVD/ ered significant). Standard of reference was a consensus evaluation of CT from MVA (p ≥ 0.796). three readers not involved in the analysis. Conclusion: The study shows that, in PDAC, CT perfusion derived BF and Results: The ICC for NCCN and KICS was 0.82 and 0.84, respectively (very IVIM derived f similarly reflect microvascularity and could possibly serve as strong agreement). The percentages of correct tumour classification at NCCN imaging biomarkers for tumor characterization. and KICS were 53-83% and 27-57%, respectively, with no statistically signifi - cant differences in the overall reader comparison per classification system. SS 8.6 Conclusion: Interreader agreement at both PC classification systems is very Relationship between pathologic treatment response strong. NCCN may be advantageous in terms of reader performance com- after neoadjuvant therapy and metabolic response on pared to KICS. integrated time-of-flight fludeoxyglucose (FDG) positron-emission tomography/MRI in patients with SS 8.4 FDG-avid borderline resectable pancreas cancer: a Intravoxel incoherent motion diffusion-weighted MR feasibility study imaging: reproducibility and diagnostic value for N. Samreen, I. Garg, J.G. Fletcher, M. Truty, G.B. Johnson, characterization of solid pancreatic lesions J.L. Fidler, K. Bradley, A. Goenka; Rochester, MN/US 1 2 1 R. De Robertis , N. Cardobi , P. Tinazzi Martini , Purpose: To determine if fludeoxyglucose (FDG) positron-emission tomogra - 3 4 4 1 M. Zanirato , A. Stemmer , R. Grimm , M. D‘Onofrio ; phy (PET)/MRI can be used to predict pathologic treatment response after 1 2 3 4 Verona/IT, Peschiera del Garda/IT, Milan/IT, Erlangen/DE neoadjuvant therapy in FDG-avid borderline resectable pancreatic ductal ad- Purpose: To evaluate the reproducibility and the diagnostic potential of intra- enocarcinoma (PDAC). voxel incoherent motion (IVIM)-derived parameters for differentiation of solid Material and methods: Patients with FDG-avid PDAC on baseline integrated pancreatic lesions and normal pancreas. time-of-flight PET/MRI who also underwent a post-neoadjuvant therapy PET/ Material and methods: Forty-seven patients with solid pancreatic tumors (30 MRI prior to surgical resection were included. Primary tumor SUVmax, SU- ductal adenocarcinomas (PDACs), and 17 neuroendocrine neoplasms (panN- Vmean and volumetric PET parameters on pre- and post-neoadjuvant therapy ENs)), 5 patients with mass-forming pancreatitis (MFP), and 30 subjects with scans were measured using anatomic guidance from simultaneously acquired normal pancreas (NP) were included. All subjects underwent 1.5 T MR imaging contrast-enhanced MRI. Metabolic response on PET/MRI was correlated to including IVIM diffusion-weighted imaging with 11 b values (from 0 to 800 sec/ histologic treatment response using College of American Pathologists grading mm ). Apparent diffusion coefficient (ADC), true diffusion (D), pseudodiffusion system (path grade). Complete metabolic response was defined as FDG up - (D*), and perfusion fraction (f) were calculated by two independent readers. take indistinguishable from surrounding background and normalization of Interobserver reliability of the measurements was assessed by using the intra- post-therapy CA 19-9 were evaluated as surrogates of path grade 1/0. class correlation coefficient (ICC). A Kruskal-Wallis H test was used for com - Results: 12 patients (range 55-79years; 50% males) underwent required PET/ parison. The diagnostic performance of each parameter was evaluated using MRI, neoadjuvant therapy, and surgical resection. Treatment response grades receiver operating characteristic (ROC) analysis. were none (grade 3, n=2), moderate (grade 2, n=6), marked (grade 1, n=2) and Results: Interobserver agreement was excellent for pancreatic lesions (ICC = complete pathologic response (pCR) (grade 0, n=2). There was no significant .981, .985, .963, and .989 for ADC, D, D* and f, respectively). ADC was not difference (p>0.05) in baseline PET parameters between group 1 (grade 3/2) significantly different between lesions. F was significantly higher in NP and and group 2 (grade 1/0). Complete metabolic response on post-therapy PET/ panNENs compared with PDACs (both p<.001). D was significantly higher in MRI was observed in 5 patients – one with grade 2, and two each with grades PDACs compared with NP and panNENs (p<.001 and .001). D* was signifi - 1 and 0. Positive and negative predictive values of complete metabolic re- cantly lower in MFP compared with panNENs, in PDACs compared with NP, sponse for path grade 1/0 were 80% and 100% whereas for normalization of and in PDACs compared with panNENs (p=.047, .047, and <.001, respective- post-therapy CA 19-9 were 20% and 33%, respectively. ly). For the differentiation between PDACs and panNENs, f had the highest Conclusion: FDG PET/MRI shows promise for response evaluation following area under the curve (.976) in ROC analysis. neoadjuvant therapy with complete metabolic response by PET/MRI correlat- Conclusion: The measurement of IVIM-derived parameters in pancreatic le- ing with marked or complete pathologic response. sions has excellent reproducibility. IVIM-related parameters could be helpful in distinguishing PDACs from panNENs. S678 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 8.7 SS 8.9 MRI imaging of IPMN: evaluation of agreement between Percutaneous radiofrequency ablation of pancreatic observers with different degrees of experience adenocarcinoma 1 2 2 2 1 1 2 1 L. Bertuzzo , G.A. Zamboni , G. Cardano , R. Pozzi Mucelli , A. Sarno , G. Tedesco , R. De Robertis , S. Paiella , 2 1 2 1 2 2 1 1 G. Mansueto ; Santorso-VI/IT, Verona/IT R. Salvia , I. Frigerio , R. Girelli , D. Melisi , C. Bassi , 1 1 2 M. D‘Onofrio ; Verona/IT, Peschiera del Garda/IT Purpose: The Sendai and Fukuoka consensus criteria were established to as- sess the risk of malignancy in pancreatic mucinous cystic lesions. The purpose Purpose: The objective of this study is to evaluate the feasibility and safety of of this study was to assess the agreement between readers with different ex- percutaneous radiofrequency ablation (RFA) of locally advanced pancreatic perience in recognizing malignancy features in a series of IPMN. cancer located in the pancreatic body. Material and methods: Institutional review board (IRB) approval was waived Material and methods: Patients with biopsy-proven locally advanced pancre- for this retrospective study. We included 118 patients (45 M, 73 F, average age atic adenocarcinoma were considered for percutaneous radiofrequency abla- 68 years) with a diagnosis of IPMN who underwent MRI/MRCP in our center. tion. Postprocedural CT studies and CA 19.9 tumor marker evaluation were Two readers, respectively with 10 and 4 years of experience in abdominal im- performed at 24 hours and 1 month. At CT, the treatment effect was evaluated aging, reviewed independently the scans assessing for wall/internal septa by excluding the presence of complications. The technical success of the pro- thickening, mural nodules, dilation of the main pancreatic duct (MPD) and con- cedure is defined at CT as the achievement of tumoral ablated area. trast enhancement. The readers were blinded to clinical and laboratory infor- Results: Thirty-five patients have been included in the study. Five of the 35 mation. Kappa statistics was calculated. patients were excluded. At CT, the mean size of the intralesional postablation Results: 10 patients had mixed-type IPMN and 108 branch-duct IPMN. Pa- necrotic area was 32 mm (range: 15-65 mm). None of the patients developed tients had a median of 2 cysts >5 mm each (range 1-20; mean 2.6). A total of postprocedural complications. Mean CA 19.9 serum levels 1 day before, 1 day 307 cystic lesions were reviewed, with a mean size of 11.8 mm (5-50 mm). The after, and 1 month after the procedure were 285.8 U/mL (range: 16.6-942.0 U/ two readers showed very good agreement regarding MPD dilation (k=0.908), mL), 635.2 U/mL (range: 17.9-3368.0 U/mL), and 336.0 U/mL (range: 7.0- wall thickening (k=0.893), and presence of mural nodules (k=0.856). The 1400.0 U/mL), respectively. The mean survival after RFA procedure of the pa- agreement was good for presence of filling defects (k=0.721), wall enhance - tients, calculated on the data collected for 26 subjects, is 312 days (range: ment (k=0.725) and presence of mural nodules (k=0.663). 65-718 days). Conclusion: Readers with different levels of experience in abdominal imaging, Conclusion: Percutaneous radiofrequency ablation of locally advanced ade- when assessing IPMN at MRI-MRCP, show an interobserver agreement rang- nocarcinoma has a high technical success rate and is effective in cytoreduc- ing between substantial (0.663) and almost perfect (0.908) when applying the tion. Fukuoka criteria, confirming their usefulness for lesion evaluation standardiza - tion. SS 8.10 Application of Fukuoka radiological criteria in SS 8.8 pathologically confirmed IPMNs of the pancreas: is it Functional imaging of pancreatic adenocarcinoma using possible to differentiate between invasive/high- perfusion CT and diffusion-weighted imaging: moderate grade dysplasia IPMNs and noninvasive/ correlation with clinicopathological features low-grade dysplasia IPMNs? J. Kovac, A. Đuri ć-Stefanović, L. Lazić, T. Nikolić, S. Jevtić, C. Fabris, G.A. Zamboni, R. Negrelli, L. Bertuzzo, D.V. Vasin, D. Masulovic; Belgrade/RS G. Marchegiani, G. Mansueto; Verona/IT Purpose: To determine the value of perfusion CT and diffusion-weighted im- Purpose: IPMNs are intraductal cystic neoplasms with malignant potential. aging (DWI) in the evaluation of patients with pancreatic adenocarcinoma, and The Fukuoka criteria are commonly used to assess the risk of malignancy. Our to assess the correlation of perfusion CT parameters, and apparent diffusion purpose was to try to differentiate between high-moderate grade dysplasia coefficient (ADC) with clinicopathological features. IPMNs and low-grade dysplasia IPMNs using the mentioned criteria. Material and methods: Forty-four patients with histologically proven pancre- Material and methods: We reviewed the preoperative MRIs from 46 patients atic adenocarcinoma who underwent CT perfusion and DWI were prospec- (27 males, 19 females; mean age 63,7 years) with a confirmed pathologic diag - tively included in the study. Tumor blood flow (BF) and blood volume (BV) were nosis of IPMN. Two readers analyzed in consensus the MRIs for tumor size automatically calculated by commercial maximum slope based software. ADC (>/< 3 cm), diameter of the main pancreatic duct (MPD) (>/< 10 mm), presence was calculated for b 0, and 800 s/mm . BV, BF, and ADC values were com- of enhancing solid components, radiologic signs of pancreatitis, enhancing or pared using Mann-Whitney test between patients with pancreatic carcinoma non-enhancing nodules, thickened/enhancing walls, abrupt change in MPD and control group. Moreover, patients with pancreatic adenocarcinoma were caliber with distal atrophy, lymphadenopathy. The data were compared using classified into two groups according to the tumor grade (30 high-grade, and 14 Fisher’s test dividing the patients into two groups: high-moderate grade dys- low-grade lesions) and the difference of perfusion CT parameters and ADC plasia IPMNs (32) vs low-grade dysplasia IPMNs (14). was assessed. Correlation was tested using Spearman’s test. Results: MPD >10 mm (p=0.04), change in MPD caliber and distal atrophy -3 Results: The mean values of BV (ml/100g), BF (ml/min/100g), and ADC (x10 (p=0.037) and radiological signs of pancreatitis (p=0.008) were significantly mm /s) were significantly lower in pancreatic carcinoma (BV=4.41±1.84, more common in patients with high-moderate grade dysplasia IPMNs than in BF=24.03±12.81, ADC=1.028±0.19) than corresponding values of healthy patients with low-grade dysplasia. Enhancing solid components were present pancreatic tissue (BV=16.92±6.88, BF=68.71±11.13, ADC=2.312±0.532). A in 12/32 high-moderate-grade IPMNs and 1/14 low-grade IPMN (p=0.07). No statistically significant difference was found for BV, BF, and ADC between high significant differences were observed between the two groups for the other and low-grade tumors (all p˂0.05). Significant correlation was found among parameters evaluated. tumor diameter and BV (ρ=-0.445, p=0.029), M-stage and BV (ρ=-0.286, Conclusion: Fukuoka criteria appear to be helpful in differentiating between p=0.049), M-stage and ADC (ρ=0.274, p=0.038), ADC and BV (ρ=0.551, invasive/high-moderate grade IPMNs and low-grade IPMNs. The most useful p=0.012), ADC and BF (ρ=0.665, p=0.023). criteria are MPD >10 mm, change in MPD caliber and distal atrophy and radio- Conclusion: Perfusion CT and DWI provide an assessment of tumor vascular- logical signs of pancreatitis. ity and cellularity and can be used for preoperative prediction of high-grade pancreatic adenocarcinoma. S679 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 Liffey Hall 1 SS 9.2 Neurons vs neural networks: comparison of machine Scientific Session SS 9 learning/computational neural networks to radiologists Radiomics and artificial intelligence in in characterization of subcentimeter liver nodules in colorectal carcinoma found on staging CT scans abdominal imaging K. Khalili, R. Lawlor, M. Pourafkari, T.K. Kim, H.-J. Jang, M. Atri, S. Johnson, A. Martel; Toronto, ON/CA SS 9.1 Purpose: To determine if machine learning could improve assessment of “too Putting things in perspective: advanced image small to characterize” hepatic nodules in patients presenting with colorectal processing with Radiomics versus the radiologist’s “gut carcinoma (CRC). feeling” to predict treatment response in rectal cancer 1 1 1 Material and methods: Imaging of patients with CRC was assessed for the J.J.M. Van Griethuysen , M.J. Lahaye , M. Maas , 2 3 4 following inclusion criteria: local CT scan in portovenous phase, <1cm hepatic H.J.W.L. Aerts , F.C.H. Bakers , R.F.A. Vliegen , 1 1 1 nodule(s) & >2-year follow-up CT/MRI. 199 patients with 547 nodules were R.G.H. Beets-Tan , D.M.J. Lambregts ; Amsterdam/NL, 2 3 4 included. Malignancy was confirmed by ≥30% change, benignity by stability of Boston, MA/US, Maastricht/NL, Heerlen/NL >2 years. Machine learning (ML): predictive features from a pool of 100 ex- Purpose: To compare the predictive performance of advanced image pro- tracted texture features were used in a random forest classifier (200 trees, cessing (Radiomics) and visual morphologic evaluation by expert-radiologists 5-fold cross-validation) to develop diagnostic models from a training set of 150 to predict response to neoadjuvant chemoradiotherapy (nCRT) on pre-treat- patients (400 nodules). In addition, five convolutional neural networks (CNN) ment MRI of rectal cancer. were trained using a fivefold cross-validation technique. Results were aver - Material and methods: The pre-treatment MRIs (T2W + b1000/1100-DWI) of aged across folds for both ML & CNN. The remaining 147 nodules were used 130 patients were analyzed to predict the likelihood of a good (Mandard tu- as the test set to compare the performance of 3 abdominal radiologists to the mour regression grade (TRG) 1-2) or complete (TRG1/pCR) response, respec- better of ML versus CNN techniques, measured through receiver operating tively, according to two methods. First, two expert-radiologists estimated the characteristic (ROC) analysis. likelihood of a good or complete response, respectively, using a 5-point confi - Results: Prevalence of disease were 25.5% and 40.5% in the learning and dence score based on an overall visual morphologic assessment (size/shape/ testing sets. In the ROC analysis of training set, CNN (mean AUC 0.89, range border/signal, TN-stage, fascia invasion, extramural vascular invasion (EMVI)). 0.84-0.94) outperformed ML (mean AUC 0.78, range 0.69-0.83) and thus was Second, 3421 Radiomics-features were extracted using PyRadiomics. Data selected for comparison to radiologists. ROC analysis of the test set demon- were analyzed using bootstrap with 100 iterations, for each iteration, patients strated a significantly better performance for radiologists (mean AUC 0.94, were randomly split 70:30 (training:testing). In the training set, top 25 features range 0.91-0.95) than CNN (0.85, p=0.008). Using a threshold of 0.5, CNN were selected using minimum-Redundancy Maximum-Relevance (mRMR). obtained a sensitivity/specificity of 80.0%/78.4%. Selected features and the radiologists‘ scores were then fitted into separate Conclusion: CNN shows promise in the characterization of subcentimeter he- logistic regression models, which were evaluated on the testing set. patic nodules in CRC though not matching expert radiologists. Future work Results: 63 patients had a good response and 29 a complete response. To with larger training set is needed. predict a good response, average area under the ROC-curve (AUC) for the two expert-radiologists was 0.69 and 0.63, versus a maximum AUC 0.66 for the best-performing Radiomics-feature (ADC LoG-5mm-GLCM-IDN; a measure of heterogeneity). To predict a complete response AUC was 0.79 and 0.69 for the expert-radiologists versus AUC 0.72 for the best Radiomics-feature (ADC LoG-5mm-GLCM-IDN). Conclusion: Visual morphologic assessment of pre-treatment MRI by expert- radiologists results in comparable performance as (individual) Radiomics-fea- tures to predict response to neoadjuvant treatment in rectal cancer. S680 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 9.3 SS 9.4 Decreased expression of OATP8 in hepatocellular Radiomics as a novel tool for primary nodal staging in carcinoma evaluated by whole-lesion radiomics on rectal cancer 1 1 1 Gd-EOB-DTPA enhanced MRI J.J.M. Van Griethuysen , D.M.J. Lambregts , S. Trebeschi , 1 1 1 2 L. Cao, J. Chen, H. Jiang, B. Song; Chengdu/CN M. Maas , M.J. Lahaye , G.L. Beets , F.C.H. Bakers , 1 3 1 R.G.H. Beets-Tan , H.J.W.L. Aerts ; Amsterdam/NL, Purpose: To investigate the value of whole-tumor radiomic features on mul- 2 3 Maastricht/NL, Boston, MA/US tiphasic gadolinium-ethoxybenzyl-diethylenetriaminepentaacetic acid (Gd- EOB-DTPA) enhanced MRI in evaluating organic anion transporting polypep- Purpose: To assess the potential of quantitative MR-imaging analysis using tide 8 (OATP8) expression in HCC patients. Radiomics for nodal characterization in primary rectal cancer staging. Material and methods: 67 consecutive surgically confirmed HCC patients Material and methods: 226 lymph nodes (from 34 rectal cancer patients who were prospectively enrolled and underwent Gd-EOB-DTPA enhanced MR ex- underwent MRI followed by surgical resection) were analyzed and matched amination. Radiomic features on each sequence were derived using in-house node-by-node with histopathology. Short axis was measured for each node software (Analysis Kit, GE Healthcare). The expression intensities of OATP8 (on T2W-MRI) and nodes were manually segmented on T1W and T2W-MRI to were histopathologically determined and scored as: 0, no expression; 1, weak extract 748 radiomic features (using PyRadiomics). Using principal feature se- expression; 2, moderate expression; and 3, high expression. Lasso logistic lection, we selected the 25 features providing best complementary information regression was used for feature selection. Correlation analysis and nonpara- to routine nodal size-measurements. Performance to distinguish between ma- metric test were conducted accordingly. Receiver operating characteristic lignant and benign nodes was assessed with areas under the curve (AUC) (ROC) analysis was applied to determine diagnostic performances of selected per-feature using a binomial mixed effects logistic regression model, correct- features in evaluating the decreased OATP8 expression. ing for patient number. False detection rate (FDR) correction set at 10% was Results: The CorrelationAllDirection_offset1_SD, IDMAllDirection_offset4_SD applied to correct for multiple testing. and LongRunEmphasis_angle45_offset1 at T1-pre imaging, the IDMAllDirec- Results: 17/226 nodes were malignant. Selected features were compared to tion_offset1_S at portal venous phase (PVP), and the IDMAllDirection_offset1_ (and combined with) nodal size-measurements using a bootstrap method with SD and IDMAllDirection_offset4_SD at hepatobiliary phase (HBP) of HCC 100 iterations; the model was trained on a random subset of patients and showed significant correlation with OATP8 expression (p<0.05). All features tested on the remaining patients (train:test ratio 80:20). Average performance differed significantly between decreased expression (score 0-2) and high ex - for nodal size-measurements was AUC 0.82. The best performing Radiomic- pression (score 3) group (P<0.05 for all). The largest area under ROC curve feature after FDR correction (gray level co-occurrence matrix (GLCM)-differ- (AUC) of T1-pre images (IDMAllDirection_offset4_SD), PVP (IDMAllDirection_ ence average; a measure of heterogeneity) resulted in an AUC of 0.83; the offset1_SD), and T1-HBP (IDMomentAllDirection_offset4_SD) were combination of this feature with nodal size resulted in an AUC of 0.85. 0.69(CI[0.55, 0.81]), 0.66(CI[0.52, 0.78]) and 0.77(CI[0.63, 0.87]), respectively, Conclusion: Radiomics may provide valuable quantitative information to char- in identifying decreased OATP8 expression in HCC. T1-pre images (IDMAll- acterize rectal cancer lymph nodes; particularly features reflecting nodal het - Direction_offset4_SD) showed highest sensitivity (87.0%, CI[66.4%, 97.2%] vs erogeneity are promising. Although Radiomics slightly improves staging per- 78.3% CI[56.3%, 92.5%], 65.2% CI[42.7%, 83.6%]) while T1-HBP (IDMomen- formance, the added benefit compared to routine size-based staging appears tAllDirection_offset4_SD) showed highest specificity (80.7% ,CI[62.5%, to be limited. 92.5%] vs 48.4%, CI[30.2, 66.9], 51.6%, CI[33.1%, 69.8%]). Conclusion: Whole-tumor Radiomics features based on Gd-EOB-DTPA en- SS 9.5 hanced MRI, especially the IDM on HBP images, offers a potential avenue to- Diffusion kurtosis MRI for evaluating inflammatory ward preoperative evaluation of the decreased expression of OATP8 in HCC. activity in ulcerative colitis: pilot study J. Podgorska, K. Pasicz, E. Zagórowicz, B. Gołębiewski, P. Kuś, J. Jasieniak, P. Wieszczy, A. Anysz-Grodzicka, J. Pałucki, E. Fabiszewska, W. Skrzyński, P. Kukołowicz, A. Cieszanowski; Warsaw/PL Purpose: To assess the efficacy of diffusion kurtosis imaging (DKI) in the as - sessment of inflammatory activity of ulcerative colitis (UC). Material and methods: 17 patients with UC underwent 3T magnetic reso- nance DKI (b values of 0-2000 s/mm ) within a short time (1-6 days) of endo- scopic evaluation and with no bowel preparation. The inflammatory activity of up to 5 bowel segments (rectum, sigmoid, descending, transverse, ascending colon) was graded on endoscopy with Mayo score. Apparent diffusion for non- Gaussian distribution (DK ) and apparent kurtosis coefficient (K) on DKI as well as apparent diffusion coefficient (ADC) on diffusion weighted imaging (DWI) were correlated with Mayo score. T-test was used to compare the groups. Results: In total 54 bowel segments: endoscopically inactive (n=18) and with mild to severe activity (Mayo 1-3) (n=36) were analysed. Statistically significant differences between inactive and active disease were found in DK (mean = -3 2 -3 2 2.3x10 mm /s, standard deviation (SD) = 0.61x10 mm /s and mean = -3 2 -3 2 1.81x10 mm /s, SD = 0.51 x10 mm /s respectively, p=0.003) and ADC (mean -3 2 -3 2 -3 2 = 1.3 x10 mm /s, SD = 0.27x10 mm /s and mean = 1.15x10 mm /s, SD = -3 2 0.23x10 mm /s respectively, p=0.041). No significant difference was found in K. Overlapping of parameters was less significant in DK than in ADC. Conclusion: DKI of UC correlates with disease activity might be superior to Gaussian DWI model for assessment of the inflammatory activity. SS 9.6 withdrawn by the authors S681 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 9.7 SS 9.9 Consistency of MR radiomic features on T2-weighted Multiparametric combined fluorodeoxyglucose-positron imaging of the liver emission tomography (FDG-PET)/CT and MR imaging to J. Santinha, J.M.G. Lourenço, I. Santiago, C. Matos, predict response to chemoradiotherapy in rectal cancer: N. Papanikolaou; Lisbon/PT whole tumor versus sub-volume analysis 1 2 1 N. Schurink , M. Berbee , J. Van Griethuysen , Purpose: To assess the intra-observer, inter-observer and test-retest repeat- 2 1 1 2 W.J.C. Van Elmpt , M. Maas , M.J. Lahaye , F.C.H. Bakers , ability of radiomic features on T2-weighted magnetic resonance imaging of the 1 1 1 R.G.H. Beets-Tan , D.M.J. Lambregts ; Amsterdam/NL, liver. Maastricht/NL Material and methods: Two axial 2D turbo-spin-echo T2-weighted abdominal acquisitions were acquired, with a minimum interval of 15 minutes between Purpose: To assess the individual and complementary performance of quanti- acquisitions, in 14 different patients (8 men; mean age=59.5 years) performing tative parameters from pre-treatment multiparametric MRI and FDG-PET/CT an abdominal MRI examination, in a „scan-rescan“ design. Examinations were to predict treatment response in rectal cancer. performed in a 1.5T Ingenia Philips scanner with the following parameters: Material and methods: A pilot group (n=20) underwent multiparametric MRI slice thickness=5mm; gap=1mm; echo time=80ms; repetition time=850ms; flip (T2W + diffusion weighted imaging (DWI); b0,100,500,1000) and FDG-PET/CT angle=90º; Bandwidth=507Hz; echo train lenght=72. Two radiologists manu- before neoadjuvant treatment + surgery. Images were anatomically co-regis- ally segmented the whole liver (VOI) twice in both acquisitions from each pa- tered using rigid + non-rigid registration. Whole-tumor volumes were segment- tient. A total of 2247 radiomic features (shape, first order, glcm, glrlm, glszm, ed on T2W-MRI and transferred to the other modalities/sequences; each tu- gldm and ngtdm of the original and filtered images - exponential, logarithm, mor was additionally divided into 7mm sub-volumes. The following parameters square, square root, two levels of wavelet and Laplacian of Gaussian with 1-,3- were calculated per-tumor and sub-volume: T2-texture (uniformity/entropy), , 5-mm) were extracted from each VOI using the Pyradiomics package. The T2-signal intensity (SI), apparent diffusion coefficient (ADC) (mean/max/min/ th th intra-observer, inter-observer, and test-retest repeatability were assessed us- median/10 & 90 percentile), standard uptake value (SUV) (mean/max/medi- ing the intraclass correlation coefficient (ICC). An ICC≥0.81 was considered an), CT-Hounsfield units (HU). Performance to predict poor response (=Mand - excellent. ard tumor regression grade (TRG)3-5) vs. good response (TRG1-2) was calcu- Results: 1084 features for Reader 1 and 1053 features for Reader 2 showed lated using receiver operating characteristic (ROC)-analysis for [1] each excellent intra-observer repeatability (mean=47.5%) with 1017 (mean=95.2%, individual parameter (whole-tumor), [2] multiparametric combination of the 5 45.3% of total) overlapping features. 1090 (48.5%) features showed excellent best-performing parameters (whole-tumor), and [3] the proportion (%) of inter-observer repeatability and 1258 (56.0%) features showed excellent test- ‘poor-response’ sub-volumes within the tumor. Sub-volumes were defined as retest repeatability. A total of 930 (41.4%) features (wavelet filters and origi - ‘poor response’ when ≥4 (out of 5) parameters within that sub-volume were nal≥100, firstorder=192 and glcm=241) showed excellent repeatability for in - indicative of TRG3-5, using a cut-off derived from dichotomisation by median tra-observer, inter-observer and test-retest repeatability. split. Conclusion: Our results indicate that, for T2-based whole-liver analysis and Results: Best single predictive parameters were T2-uniformity (area under after accounting for intra-observer, inter-observer and test-retest variability, curve (AUC)0.78), T2-SI (AUC0.70), ADCmax (AUC0.76), SUVmax (AUC0.78) less than half of radiomics features preserve an excellent repeatability. These and SUVmean (AUC0.73). Combined multiparametric performance (whole-tu- „stable“ features may, therefore, be more „trustworthy“ for T2-based whole- mor) was AUC0.91. The proportion of ‘poor-response’ sub-volumes resulted in liver radiomics analysis. AUC0.74. Conclusion: Multiparametric analysis of quantitative MRI and FDG-PET/CT data has potential added value to predict response to neoadjuvant treatment. SS 9.8 If there is a potential benefit for performing sub-volume (or voxel-wise) analy - Texture analysis of preoperative CT images for sis, this needs to be established by further and larger studies; with our current prediction of liver volume regeneration after major method, we have so far not demonstrated a clear added value. hepatectomy in living donor transplantation 1 1 1 2 1 J.E. Kim , J.H. Kim , S.J.H. Park , S.-Y. Choi , J.S. Bae , 1 1 1 2 S.J. Jeon , J.K. Han ; Seoul/KR, Bucheon/KR Purpose: To predict the rate of liver regeneration after donor hepatectomy us- ing preoperative laboratory test and computed tomography (CT) texture analy- sis in living donor transplantation. Material and methods: 112 living donors (M:F = 79:33, 31 years) who per- formed preoperative CT were included in this study. After transplantation, fol- low-up CT was performed after median 123 days. We measured the volume of future remnant liver (FLR) on preoperative CT and volume of the remnant liver (LR) on follow-up CT. Regeneration index (RI) was calculated using following equation: [(V -V )/V ] × 100. Texture analysis regarding FLR was semiauto- LR FLR FLR matically performed. We used a stepwise multivariable regression to assess associations of laboratory results and texture parameters in relation to RI and make the best fit predictive model. Results: The mean RI was 110.7 ± 37.8%, highly variable ranging from 22.4% to 247.0%. Among texture parameters, standard deviation, variance, volume of FLR and grey level co-occurrence matrices (GLCM) contrast were found to have significant correlations between RI in univariable analyses. In a multivari - able analysis adjusting for other factors, a smaller volume of FLR (ß -0.17, 95% CI -0.22 to -0.13) and lower GLCM contrast (ß -1.87, 95% CI -3.64 to -0.10) were associated with higher RI. The regression equation predicting RI was fol- lowing: RI = 203.82 + 10.42 x preoperative serum total bilirubin (mg/dL) – 0.17 x V (cm ) – 1.87 x GLCM contrast (×100). FLR Conclusion: In our study, the volume of FLR and GLCM contrast were inde- pendently associated with the rate of liver regeneration. CT texture analysis can be useful in predicting liver regeneration for the donors undergoing right hepatectomy. S682 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 Wicklow Hall 2A SS 9.10 Academic-industry collaborations involving quantitative Scientific Session SS 10 imaging biomarkers in drug development clinical trials M.S. Middleton, J. Cui, W.C. Henderson, J.C. Hooker, Liver: focal and diffuse liver pathologies, G. Hamilton, N. Szeverenyi, C. Sirlin; La Jolla, CA/US intervention and transplantation Purpose: To describe an academic-industry collaboration paradigm devel- oped over ten years while acting as a Radiology Coordinating Center (RCC) for SS 10.1 drug development clinical trials involving quantitative imaging biomarkers Gadoxetic acid-enhanced MRI before primary liver (QIBs) such as MRI hepatic proton density fat fraction (PDFF), magnetic reso- transplantation to predict recurrence of HCC nance elastography (MRE) liver stiffness, and magnetic resonance spectros- S. Lee, K.W. Kim, W.K. Jeong; Seoul/KR copy (MRS) fat tissue spectral characteristics. Our role as RRC for these com- Purpose: We performed a multicenter retrospective survey to investigate the plex, collaborative studies required iterative refinement of project planning, prognostic value of preoperative gadoxetic acid-enhanced MRI in predicting workflow, and management. HCC recurrence after liver transplantation (LT). Material and methods: We describe non-disclosure agreements and con- Material and methods: From two large LT centers, data was collected from tracts, statements of work and budgets, regulatory issues, conflicts of interest, 140 recipients with HCC who underwent primary LT and pretransplant gadox- site selection and training, intake quality control (QC), analysis, analysis QC, etic acid-enhanced MRI between January 2009 and December 2013. results reporting, protocol deviations and violations, and standard operating Results: The 1-, 3-, and 5-year recurrence-free survival rates were 91.3%, procedures. 85.4%, and 82.4%, respectively. Multivariate analysis revealed that being out- Results: We have acted as an RCC for 21 industry clinical trials involving liver side the Milan criteria (HR, 3.062; 95% CI, 1.309–7.159; P = 0.010) and peritu- QIBs, and four industry clinical trials involving MRS of the abdomen, thigh, and moral hypointensity on hepatobiliary phase (HBP) (HR, 10.268; 95% CI, 4.248– lower leg muscles. These studies were conducted at over 400 sites worldwide 24.820; P < 0.001) were independent predictors of HCC recurrence. The involving over 5,000 MR exams using the five main MR manufacturers for accuracy in categorization of the Milan criteria on pretransplant MRI was PDFF and the three MR manufacturers with MRE capability. Collectively, these 92.9% in correlation with the explanted liver. Peritumoral hypointensity on HBP exams have involved almost every conceivable combination of scanner type was significantly associated with worse tumor grade and microvascular inva - and sequence parameter settings appropriate for the measured QIBs. We have sion (P = 0.038 and P < 0.001, respectively). Using a combination of the Milan developed robust templates for an imaging manual, an RCC charter, and billing criteria and peritumoral hypointensity on HBP, preoperative MRI further strati- log and results-reporting spreadsheets. fied the risk of tumor recurrence after LT (P < 0.001), and 5-year recurrence- Conclusion: The methods, templates, and other study materials developed in free survival rates for patients outside of the Milan criteria and non-peritumoral collaboration with the pharmaceutical industry have helped us plan and man- hypointensity on HBP were comparable to those within the Milan criteria age new industry studies involving QIBs. Many of these concepts may be gen- (89.0% vs. 75.9%) (P = 0.090). eralizable to other core labs and to other clinical trials using quantitative bio- Conclusion: Pre-transplant MRI using gadoxetic acid offers not only precise markers, within and outside the field of radiology. morphologic tumor stage, but also additional biologic tumor behavior on the risk of HCC recurrence after LT. Patients outside of the Milan criteria with non- peritumoral hypointensity on HBP may achieve acceptable outcomes compa- rable to that of patients within Milan criteria. SS 10.2 Radiological-pathological correlation of pre-transplant imaging and explanted livers: American College of Radiology (ACR) Liver Imaging Reporting and Data System (LI-RADS) 2017 category versus Organ Procurement and Transplantation Network (OPTN) class in the diagnosis of HCC C.M. O Brien, A.C. O‘Brien, P.A. McCormick, N. Nolan, S.J. Skehan, R. Ryan, D.E. Malone, R. Gibney; Dublin/IE Purpose: To evaluate concordance of the LI-RADS 2017 and OPTN-United Network for Organ Sharing (UNOS) systems in the diagnosis of hepatocellular carcinoma (HCC). Material and methods: Liver transplant (OLT) patients with a pre-operative diagnosis of HCC and explants with HCC on pathology from 01/01/2014 to 09/09/2017 were retrospectively identified using the OLT database. Pre-OLT CT/MRI reports were reviewed. In patients with LI-RADS 5, 4/5, 4 or 3 lesions, the descriptions in the reports were used to categorize the lesions using the LI-RADS 2017 (LR) and OPTN-UNOS systems and the histopathology coding of explanted livers was reviewed. Data were collated in an MSExcel spread- sheet and analysed on a per patient basis. Results: 49 OLT patients met selection criteria. LR and OPTN scores were assigned to all patients. 39/49 with HCC were LR-5. Of these, 38/49 were OPTN-5, 1 did not meet arterial enhancement criteria and in retrospect was an LR-4 lesion. OPTN subclasses were: 5B 16; 5A 9; 5A-g 4; 5X 2; 5T 7. LR-4 or LR-4/5 nodules were present in 5/49 and 1/49 patients with proven HCC. 3/6 had prior TACE and were OPTN-5T. 3/6 were not OPTN 5, all patients had HCC. 4/49 were LR-3; none were OPTN 5. Conclusion: No discrepancies were found between the 2017 ACR LI-RADs 5 and OPTN-UNOS 5 lesion classification systems in these HCC patients. Me - ticulous adherence to technical and interpretative aspects of both systems is essential. An advantage of LI-RADS vs OPTN-UNOS is that LI-RADS 4 (prob- able HCC) lesions not meeting OPTN 5 criteria are identified for close follow- up. S683 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 10.3 SS 10.5 The relationship between the location of the portal vein Contrast-enhanced oncologic staging MRI of abdomen thrombus, the patency of the portal veins, and the and pelvis with a hepatocyte-specific agent in a large- coexisting disease bore MR scanner: optimizing extrahepatic tumor K. Ibukuro, G. Ogasawara, H. Fukuda, K. Tobe, M. Kishino; evaluation in neuroendocrine neoplasms 1 1 1 2 Tokyo/JP U. Fehrenbach , U. Fahlenkamp , V. Prasad , M. Pavel , 1 1 1 2 D. Geisel , T. Denecke ; Berlin/DE, Erlangen/DE Purpose: To clarify the relationship between the location of the portal vein thrombus, the patency of the portal vein, and the coexisting disease. Purpose: Liver-specific MR contrast agents in neuroendocrine neoplasms Material and methods: We retrospectively reviewed 48 patients with portal (NEN) have shown its advantages in evaluation of hepatic tumor burden. How- vein thrombus observed on contrast-enhanced CT scans for six years. The ever, a standard scan protocol is insufficient in extrahepatic and especially coexisting diseases were classified as (A) liver cirrhosis (LC) (n=29) and (B) in - pelvic tumor evaluation. The aim of the study is to evaluate an MRI shuttle flammatory diseases (Inf) including cholangitis or cholecystitis (n=13) and he - protocol with fast 3D-T1w sequences and hepatocyte-specific contrast (Gd- patic abscess (n=6). The locations of the thrombus and the patency of the EOB) to enable imaging abdomen and pelvis in optimal first-pass contrast portal vein were classified as peripheral (second-order division of intrahepatic phases in the setting of a large-bore MRI scanner with shorter axial field of portal vein, n=18) and central (n=30), complete obstruction (n=20) and patent view. (n=28), respectively. We reviewed the presence of ascites, esophageal varices, Material and methods: 96 patients with NEN of the abdomen were scanned gastro- or spleno-renal shunt, and Child classification. The logistic regression with the first pass (FP) pelvic protocol in a Gd-EOB MRI. FP pelvic sequence analysis was used to identify the statistically significant characteristics for the was acquired in between portal venous and venous liver phase. We compared location of the thrombus and the patency of the vein. The level of significance pelvic vessel and tissue enhancement between FP and a standard equilibrium was set to 5%. The Cramér’s V was measured for the association between the (EQ) pelvic sequence. Dynamic liver enhancement with the new protocol in- location, the patency, and the coexisting disease. cluding FP pelvic sequence was compared to a standard protocol without ad- Results: The coexisting disease was the primary factor that affected the loca- ditional sequence. tion of the thrombus (odds ratio: 32; LC vs. Inf) and the patency (odds ratio: 18; Results: The FP pelvic sequence showed improved vessel and tissue en- LC vs. Inf). The V were as follows: 0.693 (the coexisting disease and the loca- hancement. Aortic (140.81 vs 116.43) and venous (89.92 vs 62.76) vessel en- tion), 0.612 (the coexisting disease and the patency), 0.306 (the location and hancement in the lower abdomen showed significantly (p=0.000) higher signal the patency). intensities in FP pelvic than in the EQ phase. The FP sequence also showed Conclusion: The coexisting disease had the most important role for the loca- significantly improved lymph node enhancement (57.84 vs 41.15; p=0.000). tion of the portal vein thrombus and the patency of the portal vein. Dynamic liver enhancement in the FP protocol showed no significant differ - ences to the standard protocol (57.48 vs 53.01; p=0.143). Conclusion: Including a FP pelvic sequence in between portal venous and SS 10.4 venous liver phase by shuttling of the patients’ table enables improved tumor Assessment of liver function in patients with HCC using evaluation of the pelvis region in Gd-EOB MRI. Additionally, there was no qual- perfusion-weighted MRI with gadolinium-ethoxybenzyl- ity loss in liver enhancement in the proposed FP pelvic protocol. diethylenetriamine pentaacetic acid T. Duan, J. Chen, X. Lin, B. Song; Chengdu/CN SS 10.6 Purpose: To prospectively evaluate liver function in patients with HCC using MRI assessment of transarterial chemoembolization perfusion-weighted magnetic resonance imaging (PW-MRI) with gadolinium- effects in patients with neuroendocrine hepatic ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA). metastasis Material and methods: PW-MRI was performed on 41 HCC patients on a 3.0 M. Lapteva, M.A. Shorikov, D. Frantsev, O. Sergeeva, Tesla MR scanner. The perfusion data were acquired using a prototype radial E. Virshke, B. Dolgushin; Moscow/RU stack-of-stars 3D spoiled gradient echo pulse sequence with golden-angle radial sampling schemes over the course of 6.25 minutes. Post-processing of Purpose: The purpose of the paper is to develop suitable transarterial chem- PW-MRI data was performed on an in-house software O.K. (Omini-Kinetics). A oembolization effects (TACE) evaluation criteria in liver metastatic neuroendo- 2-compartment 2-input pharmacokinetic model was chosen to generate crine tumor (mNET) patients on MRI. haemodynamic curving. We applied an extended Toft‘s linear model to gener- Material and methods: 23 liver mNET patients (G1-9, G2-10, G3-2, unknown ate the maps of K , K , V and the semi-quantitative parameters of time-to- grade 2) who underwent 36 TACE procedures were enrolled in the study. 68 trans ep e peak (TTP), max concentration and max slope. Perfusion parameters and vol- MRI studies performed 3-86 days before (median 28 days) and 23-147 days umes of normal hepatic parenchyma were conducted, then the products of (median 61 days) after TACE were analyzed retrospectively. Using 1.5T MRI- perfusion parameters and volume were calculated. Indocyanine green reten- scanner, we acquired: 1) 3DT1WIFS in axial plane before i.v. contrast injection tion at 15 min (ICG R-15) and model for end-stage liver disease (MELD) scores and after in portal and delayed phases; 2) diffusion-weighted images with ap- were also recorded. PW-MRI parameters, volumes and products were corre- parent diffusion coefficient (ADC) maps. We measured before/after TACE: 1) lated with both ICG R-15 and MELD scores using Spearman rank correlation maximum necrotic/fibrotic zone diameter and solid component thickness on analysis. its periphery in the largest targeted lesion; 2) ADC and MR contrast agent ac- Results: The product of TTP and volume showed poor correlations with ICG cumulation dynamics (dSl) in regions of interest in tumor solid component. R-15 (rho=-0.367, p=0.046). Meanwhile, K (rho=0.496, p<0.001), K Results: TACE effect according to response evaluation criteria in solid tumors trans ep (rho=0.930, p<0.001), max slope (rho=0.342, p=0.033), product of K and (RECIST)(1.1) was evaluated as progressive disease in 8.3% cases, stabiliza- trans volume (rho=0.898, p<0.001), product of K and volume (rho=0.898, p<0.001) tion in 83.3% cases and partial response in 8.3% cases. The diameter of the ep and product of max slope and volume (rho=0.388, p=.015) correlated signifi - fibrosis/necrosis areas before/after TACE demonstrated no changes (p=0.614), cantly with MELD scores. but solid component thickness on its periphery decreased significantly Conclusion: Measures of liver function obtained by Gd-EOB-DTPA PW-MRI (p=0.003, median 16%). ADC values increased significantly in solid portions of with tracer kinetic modelling may provide a suitable method for the evaluation mNETs after TACE (p=0.003). MR contrast agent accumulation of the lesions in of liver functional reserve. portal phase decreased significantly after TACE (p=0.003, median 8.5%). De - crease of MR contrast agent accumulation in a solid part of the target lesions more than 10% after TACE is associated with a longer time to progression of the disease (p=0.025). There is an increase in MR contrast agent wash-out time in the solid part of the target lesions after TACE (p=0.032). Conclusion: Parameters changed in liver mNET after TACE: 1) solid compo- nent thickness in cases of central necrosis/fibrosis (decrease); 2) ADC (in - crease); 3) ndSI T1WI in portal phase (decrease) and this is the positive prog- nostic factor; 4) MR contrast agent wash-out time (increase). S684 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 10.7 SS 10.9 Differentiation of focal nodular hyperplasia from HCC: Gadobenate dimeglumine-enhanced MRI in patients the role of triple-phase arterial MRI with ascitic effusion 1 2 1 2 M. Gatti, R. Faletti, S. Rousset, L. Bergamasco, P. Fonio; M. Bonatti , G.A. Zamboni , F. Lombardo , R. Valletta , 1 1 2 Turin/IT G. Bonatti ; Bolzano/IT, Verona/IT Purpose: To determine the value of triple-phase arterial imaging to differenti- Purpose: To evaluate the pharmacokinetics of gadobenate dimeglumine (Gd- ate focal nodular hyperplasia (FNH) from HCC. BOPTA) in patients with ascitic effusion and to correlate it with the cause of the Material and methods: The study population included 52 patients who had effusion. arterial enhancing hepatic lesions (a total of 72 lesions: 41 HCC and 31 FNH). Material and methods: We included in our institutional review board ap- All patients underwent triple-phase arterial phase MRI using hepatocyte-spe- proved retrospective study 32 consecutive patients with ascites who under- cific agents (Gd-EOB-DTPA). Images were reviewed by two radiologists in went Gd-BOPTA-enhanced liver MRI (including hepatobiliary phase) on a 1.5T consensus: contrast enhancement ratios (CER), liver-to-lesion contrast ratios scanner in our Institution between January 2012 and December 2017. Clinical (LLC) and signal intensity (SI) were measured. The lesions were categorized data were collected. Appearance of ascitic fluid (hypo-, iso- or hyper-intense based on the peak of LLC into the following groups: 1) angiographic, 2) early in comparison to paravertebral muscle) was assessed on native and hepatobil- arterial and 3) late arterial. Data were analysed with Wilcoxon signed-rank test. iary phase gradient-recalled echo (GRE) T1-weighted images. Signal intensity Results: There was no difference in CER between FNH and HCC patients of ascites and paravertebral muscle was measured by means of a 1cm round (p>0.09). The SI increases through the different arterial phases 1) 242.08 --> region of interest (ROI) on both phases; ascites/muscle signal-to-noise ratio HCC 2) 251.84 --> 3) 253.95; however, the SI started to decrease in the late arte- (SNR) was calculated. FNH rial phase 1) 303.21 --> 2) 326.03 --> 3) 323.72. LLC were significantly Results: The cause of ascites was portal hypertension in 23/32 patients, peri- FNH higher than LLC in the angiographic phase (41.3 vs. 25.0; p=0.01), whereas toneal carcinomatosis in 6/32 and recent major surgery in 3/32. On native im- HCC there were not different in the other phases (p=0.20 and p=0.82, respectively), ages, ascitic fluid was hypointense in 90.6% of the cases and isointense in this behaviour was paralleled by the subgroup divisions: 17 (55%) FNH were 9.4%, whereas in the hepatobiliary phase it was hypointense in 3.1% of the st nd rd classified as the 1 group, 11 (35%) in the 2 and only 3 (10%) in the 3 ; on cases, isointense in 21.9% and hyperintense in 75% (p<0.001). Median as- the other hand, the HCC was homogeneously distributed (34%, 32% and cites/muscle SNR was 0.54 in native phase and 1.59 in hepatobiliary phase 34%). with a median increase of 269% (range 158-748%) (p<0.001). Median SNR Conclusion: This study highlights the different enhancement patterns of HCC increase was 294% in patients with portal hypertension, 194% in patients with and FNH: FNH resulted more conspicuous in the angiographic phase and peritoneal carcinomatosis and 239% in those after surgery (p>0.05). started its “wash out” in the late arterial phase; therefore, a multi-phase arte- Conclusion: Gd-BOPTA is excreted in ascitic fluid, which significantly increas - rial phase MRI may be an adjunctive tool to further differentiate HCC from es its signal intensity in hepatobiliary phase. No statistically significant differ - FNH. ences were found between the different types of ascites. SS 10.8 SS 10.10 Reappraisal of rate and risk factors of complications Evaluation of texture analysis for the differential after US-guided percutaneous liver biopsy: a diagnosis of focal nodular hyperplasia from retrospective analysis of 2405 biopsies hepatocellular adenoma on contrast-enhanced CT A. Maheux, S. Harguem, V. Vilgrain, M. Ronot; Clichy/FR images 1 2 2 2 R. Cannella , A. Borhani , B. Rangaswamy , M.I. Minervini , Purpose: To reappraise the rate and risk factors of complications after US- 2 2 1 2 A. Tsung , A. Furlan ; Palermo/IT, Pittsburgh, PA/US guided liver biopsy in a large single-center series of 2405 biopsies. Material and methods: We analyzed 2405 liver biopsies performed in 2137 Purpose: To investigate the value of texture analysis for the differentiation of patients (58% males, mean age 54±15 yo) between January 2008 and Decem- focal nodular hyperplasia (FNH) from hepatocellular adenoma (HCA) on con- ber 2013. Biopsies were performed for the characterization of focal liver le- trast-enhanced CT imaging. sions or to assess suspected or known chronic liver disease. Clinical, labora- Material and methods: This is a retrospective, institutional review board-ap- tory, and technical data were recorded for all biopsies. For focal liver lesions, proved study conducted in a single institution. A search of the medical records the following elements were also recorded: largest diameter, location, en- between 2008 and 2017 revealed 48 patients (47 females, 1 male) with 70 HCA hancement characteristics and pathological results. Occurrence of post-biop- and 50 patients (48 females, 2 males) with 62 FNH. All lesions were histologi- sy significant symptoms (any complaint motivating complementary imaging) cally proven and with available pre-operative CT imaging. Late hepatic arterial and complications were noted. Complications requesting specific treatment phase (HAP) and portal venous phase (PVP) images were used for texture (embolization or surgery) were considered as severe. analysis. A region of interest was placed on the largest tumor cross section Results: 1283 (53%) biopsies targeted focal lesions and 1122 (47%) were per- and in the non-lesional right hepatic lobe. Textural features were extracted us- formed for liver disease. Significant symptoms occurred after 134 biopsies ing a commercially available research software (TexRAD) which applies a (5.6%); the most frequent being pain (109/134). Complications occurred after 2-step filtration-histogram approach. The differences between textural param - 38 biopsies (1.6%); the most frequent being hemoperitoneum (n=21), subcap- eters of FNH and HCA were assessed using the Mann-Whitney U test. A p sular liver hematoma (n=11) and liver hematoma (n=6). One patient died. Se- value <0.05 was considered statistically significant. vere complications were present in 13 patients. In univariate analysis, pro- Results: On HAP images, mean, standard deviation (SD), entropy, mean of thrombin time (p=0.006), creatinine serum level (p<0.001), lesion largest positive pixels (MPP) and skewness were significantly higher in FNH than in diameter (p<0.001) and tumor pathology (p=0.040) were associated with the HCA on both unfiltered (p≤0.006) and filtered analysis (p≤0.02). On PVP mean, occurrence of complications but not platelet count. In multivariate analysis, MPP, skewness and kurtosis in FNH were significantly higher than in HCA only the lesion largest diameter was retained (OR 1.014 [1.002-1.026], (p≤0.006) on unfiltered images, while mean, SD and entropy were significantly p=0.018). higher in FNH on filtered images (p≤0.03). The analysis of the non-lesional liver Conclusion: Rate of complications after US-guided liver biopsy was low. Le- did not show any significant difference between two groups. sion largest diameter appears as the main risk factor of complication. Conclusion: Textural analysis parameters are significantly different between FNH and HCA. Imaging-based quantification of lesion heterogeneity may help in non-invasive differentiation of FNH from HCA. S685 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 The Auditorium SS 11.2 Differentiation of pancreatic neuroendocrine tumor Scientific Session SS 11 grade comparing 3D CT texture analysis and relative CT Pancreas: focal and diffuse pancreatic enhancement ratios 1 1 1 2 A. Sarno , G. Tedesco , V. Ciaravino , N. Cardobi , pathologies including endocrine tumours 2 2 1 R. De Robertis , P. Tinazzi Martini , L. Landoni , 1 1 1 1 S. Cingarlini , A. Scarpa , M. D‘Onofrio ; Verona/IT, SS 11.1 2 Peschiera del Garda/IT Additional value of tumour growth rate in patients Purpose: To evaluate the added value of 3D CT texture analysis compared to diagnosed with well-differentiated neuroendocrine CT enhancement ratios in the study of pancreatic neuroendocrine tumors tumours 1 2 3 4 (pNETs). C. Dromain , A. Sundin , P.S. Najran , H. Vidal Trueba , 5 6 7 8 Material and methods: 100 patients with pNETs were included in this study. L. De Mestier , J. Crona , M. Opalinska , L. Carvalho , 8 3 1 1 4 All tumors were pathologically diagnosed after resection or by means of bi- R. Franca , P. Borg , N. Vietti Violi , N. Schaefer , C. Lopez , 8 3 8 9 5 opsy, so histological grade was available in all cases. There were 31 G1, 52 G2, D. Pezzutti , A. Lamarca , F. Costa , M. Pavel , M. Ronot ; 1 2 3 and 17 G3 neoplasms with the presence of metastases in 53/100 (53%) cases. Lausanne/CH, Stockholm/SE, Manchester/UK, 4 5 6 7 3D CT texture analysis and a quantitative analysis of enhancement of CT stud- Santander/ES, Clichy/FR, Uppsala/SE, Krakow/PL, 8 9 ies were performed comparing the results with the tumor grading. CT texture Sao Paulo/BR, Erlangen/DE analysis results (mean value, variance, skewness, kurtosis) and relative CT en- Purpose: To assess the impact of lesion measurements and inter-reader re- hancement ratios were compared using Wilcoxon–Mann–Whitney correlation producibility on tumour growth rate (TGR) (% change in tumor volume/month) test. to predict patient outcome. Results: Included pancreatic neoplasms were located in the head of pancreas Material and methods: Baseline and 3-month (+/-1) images from patients (42%), in the body–tail (52%) and 6 patients had entire involvement of the with advanced, grade 1-2 neuroendocrine tumours (NETs) were retrospective- pancreatic gland (6%). CT texture analysis and CT quantitative analysis were ly reviewed by 2 readers for TGR calculation. Influence of number, location of 3m done in all cases. Among the three grading tumors groups (G1, G2 and G3) lesions and tumor burden on TGR on patient outcome (measured as pro- 3m kurtosis parameter resulted in statistically significant difference (p < 0.05) at CT gression-free survival (PFS)) was assessed by uni/multivariable Cox regression texture analysis. Instead, there was no statistically significant difference in CT analysis. Spearman correlation and Kruskal-Wallis test were employed. Agree- enhancement ratios between G1 and G2 groups at CT quantitative analysis. ment between readers was assessed by the Lin’s concordance coefficient Conclusion: CT texture analysis parameters can be more suitable to differenti- (LCC) and kappa (KC). ate the three grading pNETs, compared to relative CT enhancement ratios. Results: A total of 790 lesions measured in 217 patients were included. Me- dian PFS was 22.9 months. On univariable analysis, number of lesions (</=>4), SS 11.3 tumor burden and presence of liver metastases were significantly correlated to Location of liver metastases based on the site of PFS. On multivariate analysis, ≥4 lesions (HR:1.89 (95%CI:1.01-3.57)), TGR 3m primary pancreatic neuroendocrine tumour ≥0.8%/month (HR:4.01 (95%CI:2.31-6.97)) and watch-and-wait treatment cor- C. Fabris, A. Bissoli, M.C. Ambrosetti, G.A. Zamboni, related with shorter PFS. No correlation was found between TGR and num- 3m G. Mansueto; Verona/IT ber of lesions employed for TGR calculation (rho: -0.2; p value: 0.1930). No difference in mean TGR across different organs was shown (p value: 0.6). Purpose: The presence of liver metastases from neuroendocrine tumors does 3m Concordance between readers was acceptable (LCC: 0.52 (95%CI: 0.38- not preclude surgery. Knowing the distribution of these metastases is impor- 0.65); KC: 0.57; agreement: 81.55%). TGR3m remained a significant prognos - tant for treatment planning. Our purpose is to investigate whether the distribu- tic factor when data from the second reader were employed (HR: 4.35 (95%CI: tion of liver metastases from pancreatic neuroendocrine tumors (pNETs) is in- 2.44-7.79); p value<0.001) and regardless of second reader expertise (HR: fluenced by the “streamline phenomenon” within the portal vein. 1.21 (95%CI: 0.70-2.09); p value: 0.493). Material and methods: Institutional review board approval was waived for this Conclusion: Our results suggest that number of measured lesions and TGR retrospective study. We reviewed retrospectively the CTs performed on 49 pa- 3m value impact on PFS of patients with advanced NETs. The number of target tients with pNETs (27 males, 22 females) and liver metastases. Patients were lesions does not seem to affect the performance of TGR calculation. In addi- divided into 2 groups according to the pNET site: group A (11 patients) with a tion, TGR3m role as a prognostic factor was maintained regardless of reader’s pNET in the head (mean age 50.3 years); group B (38 patients) with a pNET in expertise. the body–tail (mean age 59,3 years). Two readers in consensus evaluated tu- mor site, diameter, vascular invasion and number of metastases within each lobe. Student’s and Fisher’s tests were used as appropriate. Results: Primary tumors in the body–tail (group B) were larger than those in the head (group A) (59.55 vs 39.82 mm; p=0.043). The splenic vein was more commonly involved in patients in group B. The number of liver metastases was higher in the right lobe in both groups (p=ns). The ratio of metastases in the right-to-left hemi-liver was 2.5:1 for group A compared with 1.79:1 for group B (p=0.0423). Conclusion: As expected, tumors from all locations in the pancreas will give a higher number of metastases in the right lobe. The right-to-left ratio is, how- ever, significantly different based on the site of the primary tumor. S686 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 11.4 SS 11.6 Pancreatic neuroendocrine tumor grade discrimination Pancreatic steatosis is associated with restriction of with enhancement pattern in the arterial phase: a pancreatic exocrine function 1 2 1 1 meta-analysis J.-P. Kühn , M.-L. Kromrey , R.-T. Hoffmann , M. Laniado , 2 1 2 H. Yu, M. Li; Chengdu/CN N. Friedrich ; Dresden/DE, Greifswald/DE Purpose: To assess the diagnostic test accuracy for the discriminative poten- Purpose: Recent research is now focused on the clinical significance of pan - tial of hypo-enhancement in the arterial phase to differentiate high-grade pan- creatic steatosis. The purpose was to investigate if pancreatic steatosis is as- creatic neuroendocrine tumors (HGP) (World Health Organization [WHO] grade sociated with depression of pancreatic exocrine function. II or III) from low-grade pancreatic neuroendocrine tumors (LGP) (WHO grade Material and methods: 1,458 volunteers aged 21 to 88 years (777 women) I). underwent an MRI of the pancreas, and organ fat content was quantified. Exo - Material and methods: A systematic literature search up to November 2017 crine pancreatic function was categorized by fecal elastase using defined cut- was performed to find relevant original studies. Retrieved hits were screened offs (normal function: >200 µg/g, n=1,319; moderate/manifest restriction: for inclusion and were evaluated with the revised tool for quality assessment <200µg/g, n=139). Statistical analysis to associate pancreatic fat content and for diagnostic accuracy studies (commonly known as QUADAS-2) by two re- fecal elastases includes a linear regression. Analyses were adjusted for known searchers. Pooled sensitivity, specificity, area under receiver operating charac - confounders for pancreatic steatosis, such as age sex and body mass index teristic (AUROC) curve, and other measures were calculated using random (BMI). effect models. Risk of heterogeneity was assessed for the appropriateness of Results: Overall mean (± standard deviation) of pancreatic fat content was meta-analysis. 7.50±3.78%. Significant higher levels were found in subjects with a restriction Results: Fifteen studies involving 866 patients were included. The statistical of pancreatic exocrine function (9.36±4.95%) compared to subjects with nor- heterogeneity was explained by imaging methods (CT/MR). The pooled AU- mal function (7.30±3.59%) (p<0.01). Linear regression analyses showed an ROC was 0.7494 for discrimination of HGP from LGP, with 0.632 (95% CI: inverse association between pancreatic fat and fecal elastase levels in the 0.567, 0.694) sensitivity and 0.688 (95% CI: 0.652, 0.724) specificity. whole study population [beta: -7.19 (standard error: 1.39); p<0.01] as well as in Conclusion: Hypo-enhancement in the arterial phase is a useful imaging fea- the subgroup of subjects with normal function [-4.26 (1.32); p<0.01]. Among ture to discriminate HGP from LGP. subjects with restriction in exocrine pancreatic function, a trend towards an inverse relation was detected [-1.28 (0.84); p<0.13]. Conclusion: There is an inverse relation between pancreatic fat content and SS 11.5 fecal elastases suggesting an association of pancreatic steatosis with a re- Pancreatic steatosis: quantitative analysis using MRI striction of pancreatic exocrine function. and correlation with visceral adiposity, diabetes mellitus, and chronic pancreatitis 1 1 1 2 1 T. Tirkes , T.A. Seltman , P.R. Territo , L. Li , S.A. Persohn , SS 11.7 1 3 1 1 M. Sankar , C.Y. Jeon , E.L. Fogel ; Indianapolis, IN/US, Intra-observer and inter-observer agreement on MRI 2 3 Houston, TX/US, Los Angeles, CA/US and MRCP features of chronic pancreatitis 1 2 3 4 2 T. Tirkes , A. Dasyam , Z. Shah , N. Takahashi , G. Tang , Purpose: To determine the association between the pancreatic steatosis (PS), 2 2 2 3 4 D. Yadav , K. Vipperla , P. Greer , D. Conwell , M. Topazian , visceral adiposity, diabetes mellitus and chronic pancreatitis (CP) using MRI. 1 1 2 E.L. Fogel ; Indianapolis, IN/US, Pittsburgh, PA/US, Material and methods: 118 patients were retrospectively categorized into 3 4 Columbus, OH/US, Rochester, MN/US normal (n=60), mild (n=21), moderate (n=27) and severe CP (n=10) groups based on clinical history and MRCP findings using the Cambridge classifica - Purpose: To determine intra- and inter-observer variability of MRI/MRCP for tion as the diagnostic standard. PS was calculated by measuring signal inten- evaluation of chronic pancreatitis (CP). sity on the axial water and fat-only fractions of the 2-point DIXON MR images. Material and methods: This ancillary study of the Consortium for the Study of Visceral and subcutaneous adipose tissues were separated and measured on Chronic Pancreatitis, Diabetes, and Pancreatic Cancer was performed at four axial MR image by manually outlining the anatomy using an image analyzer academic institutions. Anonymized MRI/MRCP scans of 32 patients, being software. evaluated for CP, were collected from one academic institution and reviewed Results: There was an association between the PS and distribution of the by three abdominal radiologists from other institutions. These radiologists abdominal adipose tissue, with the strongest correlation being with adiposity were asked to repeat the review two months later after scans were re-assigned within the visceral compartment (r=0.48). Higher visceral adiposity and PS a different case number. were also associated with CP; PS in the mild (24%), moderate (22%) and se- Results: Intra-observer agreement for Cambridge category varied from poor vere CP (21%) groups were significantly higher (p<0.0001) than the normal to excellent (kappa 0.33, 0.59, 0.81; concordance 47%, 69%, 84%). When group (15%). PS between the CP groups was similar (p=0.48). Smoking individual imaging findings were analyzed, intra-observer agreement was poor (p=0.003) and alcohol (p=0.02) were significant risk factors for CP. Patients to good for pancreatic atrophy (kappa 0.33, 0.66, 0.67; concordance 44%, who developed type 2 diabetes showed higher PS (p=0.01), higher visceral 53%, 66%), moderate to excellent for pancreatic duct (PD) caliber (kappa (p=0.01) and higher subcutaneous adiposity (p=0.002) as well as atrophy of 0.51, 0.65, 0.85; concordance 44%, 69%, 91%) and moderate to excellent for the tail (p=0.0009). There was a weak positive correlation between the PS and PD irregularity (kappa 0.45, 0.72, 0.89, concordance 66%, 69%, 89%). Inter- age in the normal group (r=0.33). observer for Cambridge category was moderate (kappa=0.52). Inter-observer Conclusion: Pancreatic steatosis quantitatively measured by the MRI showed agreement for individual imaging findings was poor for atrophy (kappa=0.32), a moderate positive correlation with visceral adiposity and was associated PD caliber (kappa=0.23), and PD irregularity (kappa=0.20). Patients were bal- with CP and type 2 diabetes. Prospective evaluation in well-phenotyped pa- anced in gender (50% female) with a median age of 50 years (IQR=37–58). tients is needed to confirm these results. Conclusion: There is mostly moderate, however, very variable intra- and inter- observer agreement for grading of CP using the Cambridge classification. A more robust, less subjective and more quantitative classification system, spe - cifically designed for MRI/MRCP seems necessary. This would allow improved categorization of CP patients for clinical and research purposes. S687 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 11.8 SS 11.10 MRI and MRCP features of chronic pancreatitis in MRI follow-up of intraductal papillary mucinous patients with or without pancreas divisum abnormality neoplasia: can we use a non-gadolinium-enhanced MRI and their relationship with patients’ age protocol? 1 2 2 2 M.L. Monti, G.A. Zamboni, L. Bertuzzo, R. Pozzi Mucelli, L. Bertuzzo , G.A. Zamboni , C. Fabris , R. Pozzi Mucelli , 2 1 2 G. Mansueto; Verona/IT G. Mansueto ; Santorso-VI/IT, Verona/IT Purpose: To compare the MRI and MRCP features of chronic pancreatitis (CP) Purpose: To assess whether non-contrast MRI can be safely used to follow up in patients with or without pancreas divisum (PD) and to determine their rela- patients with intraductal papillary mucinous neoplasia (IPMN). tionship with patient age. Material and methods: Institutional review board approval was waived for this Material and methods: We retrospectively reviewed 175 consecutive patients retrospective study. We evaluated 138 patients (48M, 90F, average age 64.5 (mean age 52 years) with symptomatic CP who underwent MRI and MRCP. 86 years) with an initial diagnosis of IPMN without signs of malignancy (128 had CP associated with pancreas divisum and 89 had CP without PD (biliary, branch-duct IPMN, 10 combined IPMN). All patients underwent at least two alcoholic, post-inflammatory, genetic, idiopathic). Both groups of patients consecutive MRI/MRCP with gadolinium in our center with an average follow- were stratified by age: <40 years (26 PD; 33 non-PD), 41–60 years (27 PD; 28 up of 12 months. For each patient, one reader evaluated, in separate sessions, non-PD), >60 years (33 PD; 32 non-PD). Two readers in consensus evaluated in the non-contrast and post-contrast scans: cyst size, walls/septa thickening, the scans for: morphological changes, signal intensity abnormalities, pancre- mural nodules, main pancreatic duct (MPD) dilation and contrast enhance- atic ducts changes, disease distribution. All MRI features were compared ment. across age groups using Chi-square tests. Results: We detected 376 cysts with a mean size of 10.9mm (5-29mm). Pa- Results: The pancreatic parenchyma was significantly more often T1-hypoin - tients had a median of 2 cysts each (range 1-20; mean 2.7). The MPD was di- tense in non-PD than in PD patients (p<0,0001). The degree of parenchymal lated in 22 patients, with a mean caliber of 4.9mm (4-9mm). In the first MRI thinning and irregularity of parenchymal margins was significantly higher in available, in the non-contrast scans, 356 (94.7%) lesions did not show suspi- non-PD patients in the groups aged <40 and 41–60 (p<0.0001) while no differ- cious signs. We detected walls/septa thickening in 4 cysts and endoluminal ence was observed between PD and non-PD patients aged >60 (p=ns). The defects in 16: 15 were classified as debris because of their dependent posi - main pancreatic duct (MPD) was significantly more often dilated in non-PD tion, 1 as mural nodule. After contrast administration, none of these showed patients, in all age groups (p=0.0031). Filling defects were more common in enhancement. At the next available follow-up MRI, in the non-contrast scans, non-PD patients (p<0.0007). A significant association was found between pa - 360 (95,7%) lesions did not show any changes, 11 (3%) showed a slight in- renchymal T1-hypointensity, MPD dilation and filling defects and increasing crease in size and 5 (1,3%) showed signs of progression. This was confirmed age in non-PD patients compared to PD patients in all age groups (p<0.001). in the post-contrast scans. Conclusion: Patients with PD show less severe morphological alterations than Conclusion: At follow-up, non-contrast MRI would have been enough to ex- patients affected by other forms of chronic pancreatitis; these morphological clude signs of malignant evolution in 98.7% cysts and 96.4% patients. alterations are significantly associated with increasing age. SS 11.9 Value of MRCP for the detection of intraductal papillary mucinous neoplasm of the pancreas V. Grossarth, J. Boddenberg, N. Ziayee, F. Verfürth, T. Lauenstein; Düsseldorf/DE Purpose: To assess the diagnostic value of MRCP for the detection of intra- ductal papillary neoplasm (IPMN) and to compare the results with endoscopic retrograde cholangiopancreatography (ERCP), endosonography and histopa- thology as the standard of reference. Material and methods: In this retrospective study, 98 patients with suspected IPMN were analyzed. All subjects underwent an MRI examination on a 1.5T scanner (Magnetom Avanto, Siemens) including 3D MRCP sequences. Two radiologists evaluated MRI data in a consensus mode regarding the presence of main duct (MD) or side branch (SB) IMPN. ERCP, endosonography and his- topathology were performed within two weeks of MRI and served as a stand- ard of reference. Results: Presence of IMPN was revealed in 80 of 98 patients. Overall sensitiv- ity and specificity of MRCP amounted to 97% and 62%, respectively. While sensitivity for the depiction of SB IPMN was as high as 97%, only 77% of MD IPMN was correctly identified by MRCP. Conclusion: MRCP is an accurate tool for the detection of IPMN in general. However, the diagnosis of MD IPMN should be verified in a clinical setting by further diagnostic tools since therapeutic options strongly depend on the cor- rect diagnosis. The only moderate specificity of MRCP may be related to the misdiagnosis of IPMN as other cystic pancreatic lesions. S688 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 The Liffey A SS 12.3 CTC in patients undergoing surgical management for Scientific Session SS 12 deeply infiltrating endometriosis CT colonography and colonic imaging S. Liddy, E. O‘Dwyer, R. Sarkar, A. O‘Neill, J. Feeney; Dublin/IE SS 12.1 Purpose: To determine the value of CTC in predicting the need for bowel re- Performance and evaluation in CTC screening section and stoma formation in patients undergoing surgical management for 1 2 1 A.E. Obaro , A. Plumb , R. Baldwin-Cleland , deeply infiltrating endometriosis. 1 2 1 1 C. Ugarte-Cano , S. Halligan , D. Burling ; Harrow/UK, Material and methods: A retrospective review of all patients who underwent London/UK surgical management of deeply infiltrating endometriosis at a tertiary referral centre over a 12-month period. CT colongraphy was correlated with intraop- Purpose: To implement a structured CTC training programme with individual- erative findings. Data recorded included patient demographics, medical his - ized feedback and assess radiologist attitudes to the programme and CTC tory, sites of disease involvement, the plane of contact between the bowel and quality metrics. the surrounding organs and/or endometrial deposits, the presence of luminal Material and methods: CTC reporting radiologists were randomised to either narrowing and the surgical procedure undertaken. A per-patient and per-lesion an intervention group (received one-day, 1:1 CTC training and feedback on analysis was performed using the Student’s t test, chi-square test and one- test performance) or a control group (no training or feedback). Both groups way analysis of variance. took four CTC assessments to assess detection of colorectal cancer (CRC) Results: Eighteen patients with a total of 25 sites of bowel involvement under- and polyps >5mm. Tests were administered at baseline, 1 month after work- went surgery in the 12-month period. The mean age was 39. Sites of bowel shop, 6 months and 12 months. CTC workload and views on CTC quality met- involvement included rectal (13/25), rectosigmoid (3/25), sigmoid (7/25) and rics were obtained via questionnaires. caecal (2/25). The mean plane of contact was 1.9 cm. Luminal narrowing was Results: 97 radiologists have been recruited and have a wide range of CTC present at 8/25 sites of bowel involvement. A longer plane of contact was as- reporting experience, from less than 50 cases (2 of 97, 2%) to over 3000 (10 of sociated with a need for bowel resection (2.6 vs 1.7 cm, p=0.04). Factors not 97, 10%). Most radiologists would find knowledge of their polyp detection rate associated with bowel resection included patient age (38 vs 39, p=0.44), mul- valuable or very valuable (92 of 97, 95%). 51 radiologists are randomised to the tifocal disease, site of bowel involvement and the presence of luminal narrow- intervention group. Of these, 28 have taken baseline test and attended the ing (2/7 vs 8/18, p=0.8). workshop. 71% (20 of 28) found the baseline test difficult, and all either agreed Conclusion: A longer segment of bowel involvement may be predictive of a (9 of 28, 32%) or strongly agreed (19 of 28, 68%) that the workshop provided need for bowel resection and stoma formation in patients undergoing surgical useful feedback regarding their baseline test performance as well as addi- management for deeply infiltrating endometriosis. tional learning opportunities (21 of 28, 75%, strongly agreed). 96% (27 of 28) felt motivated to independently improve their CTC performance. Conclusion: Early results demonstrate that a 1:1 CTC training workshop is SS 12.4 well received and motivates radiologists to improve their practice. Radiologists MR-enema (MR-e) and CTC in the diagnosis of recto- would find knowledge of their validated polyp detection rate very valuable. sigmoidal endometriosis: a preliminary comparative evaluation E. Biscaldi, F. Barra, C. Scala, V. Vellone, S. Ferrero; SS 12.2 Genova/IT Patient anxiety prior to a first CT examination to investigate abdominal malignancy: is CTC different from Purpose: To compare the accuracy of MR-e and CTC in detecting and evalu- standard abdominal CT? ating recto-sigmoid endometriosis. 1 2 2 1 D.J.M. Tolan , C. Roe , C.G.D. Clarke ; Harrogate/UK, Material and methods: We collected the patients following these inclusion Leeds/UK criteria: reproductive age, suspicion of deep pelvic endometriosis on the basis of symptoms and vaginal examination, presence of gastrointestinal symptoms, Purpose: This study evaluates patients undergoing CT for the first time and no previous radiological imaging. Patients underwent laparoscopy within 3 whether patients undergoing CTC have different levels of anxiety than those months from the diagnostic procedures. Radiological findings were compared having standard CT of the abdomen and pelvis (CTAP) when performed as a with surgery and histology. search for cancer. Results: Out of 85 women included, 52 had rectosigmoid endometriotic nod- Material and methods: Adult patients referred for either CTC or CTAP (single, ules (61.2%). The sensitivity, specificity, positive predictive value, negative dual or triple phase) were eligible for this pilot study. 60 consecutive eligible predictive value, likelihood ratio positive and likelihood ratio negative of MR-e patients were recruited (30 in each group). Participating patients completed and CTC in the diagnosis of recto-sigmoid endometriosis were 88.46% the state–trait anxiety inventory (STAI) questionnaire in the department prior to (95%C.I., 76.56%-95.65%), 90.0% (73.47%-97.89%), 93.88% (83.92%- CT scan. The study was granted health research authority ethics approval. 97.83%), 81.82% (67.75%-90.60%), 8.85 (3.01-26.0), 0.13 (0.06-0.27) and Results: Patients were recruited between March and October 2017. The CTC 92.31% (81.46%-97.86%), 93.75% (79.19%-99.23%), 96.0% (86.22%- group was older (CTC mean 77.7, range 63-90 vs. CTAP mean 56.6, range 25- 98.93%), 88.24% (74.44%-95.08%), 14.77 (3.85-56.64), 0.08 (0.03-0.21), re- 81, p=<.00001) and had more females (CTC n=23/30; 76.6% vs CTAP n=13/30; spectively. No significant difference was appreciated in the accuracy of the 43.3%, p=0.008). However, there was no significant difference in the baseline two techniques in the diagnosis of rectosigmoid endometriosis (p = 0.607; ‘T-anxiety’ measure (mean CTC 36.3 vs CTAP 34.6; p=0.42). The immediate McNemar’s test). CTC was more effective in evaluating the distance between pre-scan ‘S-anxiety’ was significantly higher for CTC (mean CTC 42.73 vs the nodule and the anal verge in all cases, it was less precise in evaluating the CTAP 33.73; p=0.00614). Patients having CTC have significantly higher anxi - depth of the lesions in the intestinal wall. CTC was more panoramic in whole ety levels before the scan from baseline levels (CTC mean T-anxiety 36.3, S- colon evaluation. anxiety 42.73, p=0.004296). Conclusion: CTC is competitive with MR-e in this diagnosis. CT software also Conclusion: Patients experience significant anxiety prior to CTC over those allows measurement of the distance of the nodule from the anal verge. CTC is attending for CTAP when investigating cancer. Further work is required to iden- probably still limited in evaluating the depth of penetration of the nodule in the tify the impact of anxiety on patient perception of care and effective approach- intestinal wall, due to the lack of contrast resolution. In the future, CTC may be es for managing patient anxiety in those having CTC. implemented with iodinated medium injection to boost its diagnostic yield. S689 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 12.5 SS 12.7 Incidence of clinically significant perforation at low- MRI parameters for prediction of sigmoid carcinoma in dose non-contrast CT and its value prior to same day patients with background of diverticular disease: a CTC following incomplete colonoscopy retrospective study 1 2 2 1 A. O‘Shea, M. Morrin, M. Lee, E. Thornton, T.É. Murray; E. Savoldi , A. Higginson , R. Beable , C. Ball ; 1 2 Dublin/IE Cosham/UK, Portsmouth/UK Purpose: Routine low-dose non-contrast CT of the abdomen and pelvis has Purpose: To retrospectively identify morphologic-specific findings at MRI in been recommended prior to same-day completion CTC to assess for occult the differentiation of mass-like diverticular disease from sigmoid carcinoma, in perforation at preceding incomplete colonoscopy, before insufflation at CTC. patients with background of diverticular disease and sigmoid stricture at CT. The aim of our study is to examine the incidence of clinically significant perfo - Material and methods: The cohort consisted of 37 patients with thickening/ ration at low-dose CT. We also examine the benefits of low-dose pre-scan in stricture of the sigmoid colon at CT, representing diverticular disease (n = 32) assessing adequacy of bowel preparation and identifying any other relevant or sigmoid carcinoma (n = 5). MRI studies were scored according to the pres- contraindications to CTC. ence or absence of potential discriminators. Patients underwent CT, T2- Material and methods: We conducted a retrospective review of all low-dose weighted and diffusion-weighted MRI. Sensitivity, specificity, positive predic - non-contrast CTs performed following failed colonoscopies over a one-year tive value (PPV), negative predictive value (NPV), and accuracy were calculated period (n=135). We also assessed the adequacy of bowel preparation on a and MRI were retrospectively reviewed using three different parameters: diffu- scale of 1-5, in order of increasingly adequate preparation. Incidentally noted sion restriction, loss of wall stratification and mucosal lesion. contraindications to CTC were also recorded. Results: MRI showed high NPV (93%) and specificity (96%) and PPV and Results: One perforation (which was extraperitoneal) was identified on low- sensitivity of 75% and 60%, respectively, with an overall accuracy of 91%. dose CTC. However, 6 patients were found to have contraindications to CTC Diffusion restriction had a sensitivity, specificity, and PPV of 100%, 96% and including acute diverticulitis and acute colitis. Overall, the bowel preparation 83%, respectively, for the diagnosis of cancer. Loss of wall stratification had a was felt to be adequate (≥ 3) in 96% percent of patients. 5 patients (0.04%) sensitivity, specificity, and PPV of 60%, 96%, and 75%, respectively, while the had their CTC postponed due to poor bowel preparation. presence of mucosal lesion had a sensitivity, specificity, and PPV of 80%, Conclusion: While the incidence of perforation identified at low-dose pre-scan 100%, and 100%. The analysis of the three parameters may change the MRI is low, there may be additional benefits to a low-dose non-contrast CT prior to accuracy at 97%. insufflation including assessment of adequacy of bowel preparation and iden - Conclusion: Diverticular disease is best differentiated from carcinoma by the tifying other contraindications to same-day CTC. analysis of three parameters: preservation of the wall layering, the absence of restricted diffusion and the absence of mucosal mass with low T2 signal. MRI improves the differentiation between sigmoid cancer and diverticulitis. These SS 12.6 encouraging results need to be confirmed in a larger study. 7 T MRI for fibrosis evaluation in a radiation-induced murine model of colitis 1 2 1 3 M. Zappa , S. Doblas , D. Cazals-Hatem , F. Milliat , SS 12.8 1 1 1 1 P. Garteiser , M. David , V. Vilgrain , B. Van Beers , Does the abdominal deep subcutaneous fat distribution 2 1 2 E. Ogier-Denis ; Clichy/FR, Paris/FR, influence complication rate and mortality after elective Fontenay-Aux Roses/FR surgery for colorectal cancer? E. Frostberg, Y. Manhoopi, M.R. Pedersen, H. Rahr, Purpose: Crohn‘s disease transmural bowel wall inflammation can lead to fi - S.R. Rafaelsen; Vejle/DK brosis causing luminal narrowing and fistula which are the main indications to surgery. The aim of our study was to evaluate MRI to distinguish submucosal Purpose: Intraabdominal visceral adipose tissue is associated with high risk of from transmural fibrosis using an original murine model of radiation-induced developing arteriosclerosis, and cardiovascular co-morbidity is associated colitis. with high complication and mortality rates after colorectal cancer surgery. Ac- Material and methods: Colitis was induced with localized single-dose radia- cording to the new literature, the deep subcutaneous adipose fat (dSAT) is like tion (27 Gy). We included an inflammation group of 24 rats with pathologic the intraabdominal visceral adipose tissue. The aim of this study was to evalu- features of severe acute inflammation associated with minimal submucosal fi - ate if the dSAT, measured by a regular CT scan, may be used as a predictor for brosis and a mixed group of 39 rats with pathological features of severe inflam - complications and mortality after colorectal cancer surgery. mation associated with transmural fibrosis, obtained two and twelve weeks Material and methods: Preoperative CT scans of 344 patients who under- after irradiation, respectively. Fat-suppressed T2- and T1-weighted, diffusion- went colorectal cancer surgery in the years 2010 and 2011 were examined weighted, magnetization transfer and perfusion using arterial spin labelling retrospectively. Measurements of the dSAT were collected. Clinical data re- (ASL) technique 7 T MRI was performed. garding patient characteristics, postoperative surgical and medical complica- Results: MRI showed significant differences between inflammation and mixed tions and survival data were retrieved. The radiological measurements were groups in normalized to muscle signal intensity on T2-weighted images (4.2 vs used as explanatory variables in fitted logistic and Cox regression models. 3.1, P< 0.0001) and T1-weighted images (1.4 vs 1.3, P=0.0003), ADC (2.17×10 Results: 325 patients were eligible for the study. The dSAT had a significant 3 -3 2 vs 1.69×10 mm /s, P<0.0001), magnetization transfer ratio (35 vs 42%, relationship with body mass index, and body mass index had a significant as - P<0.0001) and perfusion (60 vs 38 mL/min/100g, P=0.0009). Monoparametric sociation with surgical complications (OR: 1.09, p=0.01). Surgical complication analysis with the best area under the curve (AUC) to differentiate the two had a strong impact on medical complication (OR:5.9, p=0.0001). Union for groups included T2-weighted signal intensity, apparent diffusion coefficient international cancer control (UICC) stage, age and co-morbidity had a signifi - (ADC) and magnetization transfer ratio. The combination with the best global cant association with long-time survival. The radiological measurement of the predictive value (94%) included all parameters but T1-weighted parameter dSAT did not have an impact on surgical or medical complications, nor did it (AUC 0.94). have an association with long-time survival. Conclusion: MR multiparametric analysis was able to distinguish transmural Conclusion: Measuring the thickness of the dSAT in a preoperative setting from submucosal fibrosis in our radiation-induced model. This model could be does not have any impact on postoperative surgical or medical complications used to evaluate antifibrotic treatments currently under development. nor long-time survival in colorectal cancer patients treated with elective sur- gery. S690 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 Liffey Hall 1 SS 12.9 Model-based iterative reconstruction in ultralow-dose Scientific Session SS 13 CTC: diagnostic performance and influence on computer-assisted diagnosis Diffuse and chronic liver disease S. Barbieri, G. De Paoli Barbato, E. Raimondi, M. Bassi, A. Deledda, M. Tilli, R. Rizzati, M. Giganti, G. Benea; SS 13.1 Ferrara/IT Liver imaging findings in adult patients with Fontan Purpose: To compare the perfomance of standard dose (SD) and ultralow circulation dose (ULD) in CTC using hybrid-iterative reconstruction (HIR) and model- H.K. Yang, H.-J. Jang, R.M. Wald, S.-J. Yoo, T.K. Kim; based iterative reconstruction (MBIR), and to assess the effect of different re- Toronto, ON/CA construction algorithms on computer-aided diagnosis (CAD). Purpose: To describe liver imaging findings and complications on CT or MRI Material and methods: We evaluated 50 patients (31 males, 19 females, in adult patients with Fontan circulation and to investigate as to whether he- mean age 69.7) through CTC with SD (120kV, 50mAs 4.6±1.3mSv) and ULD patic imaging features show correlation with post-Fontan duration, age or se- (120kV, 10mAs, 0.95±0.2mSv); images were reconstructed with both HIR and rum hepatic function tests. MBIR. A commercial CAD system was applied complementary to all recon- Material and methods: Our institutional review board approved this retro- structions. Two independent radiologists reviewed all obtained images sepa- spective study with waiver of informed consent. Two blinded abdominal radi- rately and in random order to avoid recall biases. The two readers recorded ologists retrospectively reviewed the abdominal CT (n = 21) and MR (n = 16) colonic findings (polyps/cancer), extracolonic findings and CAD detections. images of 37 adults (mean age, 30 years; 14 men and 23 women) with Fontan Polyps (p) were classified by size (p<5mm, 5≤p<10mm, p≥10mm), shape (flat, circulation (mean post-Fontan duration, 23 years). Biochemical hepatic func- sessile, pedunculated) and location (rectum, sigmoid, descending, transverse, tion tests within 6 months of CT or MR examinations were used for analysis. ascending, cecum). Only relevant extracolonic findings (C-RADS: E4) were re - Results: Inhomogeneous, reticulated-mosaic pattern of hepatic parenchymal corded. CAD results were reported and significant findings were selected. enhancement, lobulated hepatic contour, blunt margin, and hypertrophy of the Results: 9 polyps were found in 8 patients from: six <5mm, two ≥5_<10mm left lateral section were seen in 86% (32/37), 96% (34/37), 95% (35/37) and and one ≥10mm; 8 were pedunculated and 1 flat 2 were in cecum, 3 ascend - 73% (27/37) of patients, respectively. Among the analyzed 25 arterial hyperen- ing, 2 transverse, 1 descending, 1 sigmoid tract; colonic-finding records were hancing nodules in 16 patients, 2 nodules (8%, 2/25) in 2 patients (13%, 2/16) superimposable for both radiologist with SD reconstructions and with ULD were hepatocellular carcinomas (HCCs) and the remaining 23 nodules in 14 MBIR reconstruction; 11 extracolonic findings E4 were recorded with SD, 2 patients were diagnosed as focal nodular hyperplasia (FNH)-like nodules. The with ULD (AAA, hiatal hernia). Colonic findings were confirmed by optical colo - diameter of inferior vena cava showed significant positive correlation with noscopy. post-Fontan duration, age and laboratory hepatic function test. Conclusion: CTC with ULD and MBIR reconstructions showed promising re- Conclusion: Imaging findings of adult patients with decades of Fontan circula - sults in terms of diagnostic performance and inter-reader reproducibility for tion were characterized by reticulated-mosaic pattern of parenchymal en- intra-colonic findings. CAD detection of significant findings was unaffected by hancement, lobulated contour, blunt margin, and hypertrophy of the left lateral acquisition protocol or reconstruction algorithm. section. Arterial-phase hyperenhancing FNH-like nodules are commonly seen, but HCC can infrequently occur. SS 12.10 Dynamic MR of the pelvic floor: impact of the SS 13.2 pubococcygeal line on the grading of pelvic floor Evaluation of two-point Dixon water-fat separation for descent gadoxetic acid-enhanced MRI-based estimation of liver 1 2 1 1 2 S. Picchia , M. Rengo , D. Bellini , M. Montesano , A. Laghi ; function 1 2 Latina/IT, Rome/IT M. Haimerl, S. Poelsterl, C. Stroszczynski, P. Wiggermann; Purpose: To evaluate the impact of three methods for drawing the pubococ- Regensburg/DE cygeal line (PCL) on the grading of pelvic floor descent with MR. Purpose: To evaluate the impact of intrahepatic fat on gadoxetic acid (Gd- Material and methods: Female patients with suspected pelvic floor dysfunc - EOB-DTPA)-enhanced T1 relaxometry for the estimation of liver function using tion were prospectively included in the study. All exams were acquired on a 1.5 T1-weighted volume-interpolated breath-hold examination (VIBE) sequence Tesla MR with patient in supine position. Rectum and vagina were filled with with Dixon water-fat separation. 200 ml and 50 ml of ultrasound gel, respectively. A distended bladder was re- Material and methods: 74 patients underwent Gd-EOB-DTPA-enhanced T1 quired. The acquisition protocol included turbo spin echo (TSE)-T2-weighted 13 13 relaxometry and C-methacetin breath test ( C-MBT) for the evaluation of sequences on the three planes for morphological evaluation. A steady-state liver function. Prior to (T1 ) and 20 minutes after the intravenous administra- pre sequence (True-FISP) in the midsagittal plane was acquired during straining tion of Gd-EOB-DTPA (T1 ), a T1-weighted VIBE sequence with Dixon water- post and defecation, acquiring 1 image/sec. The anterior point of all PCL was the fat separation and multiple flip angles (1°, 7°, and 14°) was used to generate pubic symphysis, while posterior points were the tip of the coccyx (PCLtip), both T1 maps (T1_in) of the in-phase component without fat separation and T1 the sacrococcygeal joint (PCLsc) and the last coccygeal joint (PCLcc). The maps merely based on the water component (T1_W). The reduction rates of T1 grade of pelvic floor descent was measured with the M-line according to the relaxation time (rrT1) between pre- and post-contrast images were evaluated HMO system, traced perpendicularly from the PCL to the posterior end of the 13 for T1_in (rrT1_in) and T1_W (rrT1_W). C-MBT values were correlated with T1 H-line. Measurements of M-line were compared as differences of absolute and rrT1. post measures and differences of grading. Results: A constant significant decrease of rrT1 with progressive reduction of Results: The final population consisted of 29 consecutive patients (mean age liver function could be shown for both T1_in (p < 0.001) and T1_W (p < 0.001). 56.07yy±10.52). M-line measured using PCL tip was significantly smaller than 13 When patients are subdivided into 3 different categories of C-MBT readouts, PCLsc or PCLcc in all measurements. The grading of pelvic floor relaxation, the groups can be significantly differentiated by their rrT1_in and rrT1_W val - using PCLtip, was underestimated in 13 patients (44%) at rest, in 19 patients 13 ues (p<0.005). Simple regression model showed a log-linear correlation of C- (65%) during straining and in 16 patients (55%). MBT values with T1_in (r = 0.566; p<0.001), T1_W (r = 0.568; p<0.001), post post Conclusion: A significant underestimation of the pelvic floor descent was ob - rrT1_in (r = 0.726; p < 0.001) and rrT1_W (r = 0.766; p<0.001). served when the M-line was measured using the PCLtip as reference. The use 13 Conclusion: Liver function as determined using C-MBT can be estimated of PCLsc or PCLcc should be used to quantify the pelvic floor descent. from Gd-EOB-DTPA-enhanced MR-relaxometry more accurately when intra- hepatic fat is taken into account. Here, T1_W maps are significantly superior to T1_in maps without separation of fat. S691 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 13.3 SS 13.5 Disodium gadoxetate-enhanced liver MRI in cirrhotic Assessment of liver function in patients with chronic patients: can clinical parameters predict suboptimal liver diseases using different parameters of gadoxetic hepatobiliary phase enhancement? acid-enhanced hepatobiliary phase imaging: a S.P. Walker, D. Wilson, H. Ramotar, C.R. Smith, R. Albazaz, comparison study M.B. Sheridan, J.A. Guthrie; Leeds/UK L.B. Beer, N. Bastati-Huber, S. Pötter-Lang, D.P. Stoyanova, M. Elmer, D. Tamandl, A. Ba-Ssalamah; Vienna/AT Purpose: The potential added value of disodium gadoxetate (Gd-EOB-DTPA) is its active uptake by hepatocytes to provide a hepatobiliary phase (HBP) aid- Purpose: To evaluate the correlations between different hepatobiliary phase ing detection and characterisation of focal liver lesions. Predicting suboptimal (HBP) image scores using gadoxetic-acid (GA) uptake and the albumin-biliru- HBP enhancement would be advantageous so as to select a different contrast bin (ALBI) grading system, in patients with chronic liver disease. medium saving expense and room time. This study evaluates the relationship Material and methods: This retrospective institutional review board approved of clinical parameters to HBP enhancement using Gd-EOB-DTPA in cirrhotic study included 220 patients (139 male, 81 female, mean age 54.9 ±14.3 years) patients. with chronic liver disease who had undergone GA-enhanced MRI of the liver Material and methods: Cirrhotic patients undergoing Gd-EOB-DTPA MRI be- between 2011 and 2015. Relative liver enhancement (RLE), hepatic uptake tween January 2016 and December 2016 were retrospectively reviewed. index (HUI), contrast uptake index (CUI), and liver-to-spleen contrast ratio Based on region-of-interest sampling, the liver signal intensity was calculated (LSC) were calculated using unenhanced and GA-enhanced HPB images, 20 using the spleen as a reference. The liver spleen contrast ratio (LSCR) was minutes after administration of the contrast material. ALBI grading was used then calculated and compared with basic patient characteristics and labora- as a clinical score for correlation with disease severity. Correlation coefficients tory results. between different factors were calculated using Pearson‘s correlation coeffi - Results: 127 patients were identified with an average HBP time of 19 minutes cient. 31 seconds. Twelve patients were excluded due to incomplete availability of Results: The highest correlation coefficients were observed between RLE-CEI clinical parameters. Multiple logistic regression analysis revealed that bilirubin (R=0.904, p<0.001), RLE-LSC (R = 0.782, p<0.001), and CUI–LSC (R=0.787, (p=0.04), albumin (p=0.01), PT (p=0.02) and platelets (p=0.04) were factors p<0.001). Correlation strength between HUI and RLE, CEI, and LSC was predictive of HBP enhancement. These clinical parameters were also signifi - (R=0.564, R = 0.535, R = 0.434, respectively, p<0.001 for all). R values be- cant with a simple logistic regression analysis, except for platelets (p=0.86). tween the ALBI score and RLE, CUI, HUI, and LSC were -0.481. -0.442, -0.335, Conclusion: HBP enhancement in cirrhotic patients correlates with a number -0.416, respectively, indicating a moderate correlation. of clinical biomarkers, the strongest being albumin. Prospective identification Conclusion: A strong correlation between different MR-derived parameters of suboptimal HBP enhancement using such biomarkers could avoid the non- using HBP imaging and the ALBI score, except for the HUI, was observed. beneficial use of Gd-EOB-DTPA in cirrhotic patients thus saving money and These data indicate that, except for volume-based scores, such as the HUI, resources in favour of alternative contrast medium. other MR-derived parameters are interchangeable for the evaluation of diffuse liver disease. SS 13.4 Assessing liver function: diagnostic efficacy of SS 13.6 parenchymal enhancement and liver volume ratio during Comparison of hepatic R2* and fat fraction obtained by hepatobiliary phase of gadolinium-ethoxybenzyl- pixelwise map methods to MRQuantif software results diethylenetriamine pentaacetic acid-enhanced MRI A. Paisant, C. Chargeboeuf, J. Jouan, H. Saint-Jalmes, studies E. Bardou-Jacquet, Y. Gandon; Rennes/FR A. Pecorelli, E.B. Orsini, C. Talei Franzesi, S. Famularo, Purpose: The aim was to compare the results of MRQuantif freeware using a L. Gianotti, D. Ippolito; Monza/IT generic 2D multi-echo gradient echo sequence (ME-GRE), previously validated Purpose: To assess whether gadolinium-ethoxybenzyl-diethylenetriamine against biopsy, to constructor sequences providing R2* and fat fraction (FF) pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI study is useful to determine maps. liver function in comparison to Child-Pugh (CP), model for end-stage liver dis- Material and methods: After institutional review board (IRB), also known as ease (MELD) and biochemical test. an independent ethics committee (IEC), ethical review board (ERB), or re- Material and methods: All Gd-EOB-DTPA-enhanced MRI studies performed, searchIRB approval and written informed consent, on a 3.5-year period, we between May 2011 and September 2017, to characterize focal liver lesion explored 619 patients for hepatic iron or fat quantification. For each we ob - were retrospectively reviewed. Patients were divided into study and control tained a generic 2D ME-GRE with TE multiple of 1.2 ms and a constructor 3D group according to the presence/absence of liver cirrhosis. Signal intensity ME-GRE providing a R2* or T2* and a FF maps, either Ideal-IQ (n=113) on a was calculated as the rate of liver-to-muscle ratio on contrast-enhanced T1- 1.5T MR450 (General Electric), 3D-Vibe-Dixon WIP (n=196) on a 3T Verio (Sie- GE sequences in portal (SI-POR) and hepatobiliary phase (SI-HEP) and then mens) or mDixon-Quant (n=310) on a 3T Ingenia (Philips). We then compared normalized for liver volume (SI-POR/LV and SI-HEP/LV). (correlation and Bland-Altman) R2* and FF determined by MRQuantif to the Results: A total of 303 Gd-EOB-DTPA-enhanced MRI studies, performed in value of the two maps, from the mean of 3 region of interests (ROI)s placed at 221 consecutive patients, were included. Cirrhosis was present in 191 (63.0%) the same location on the right liver. Discordant data were specifically analyzed MRI studies. SI-HEP was significantly lower in cirrhotic than in non-cirrhotic by an expert. patients (0.55±0.29 vs 0.66±0.40, p=0.004), while no difference was found in Results: After exclusion of map reconstruction miscalculation errors (FF>50%, SI-POR. SI-HEP progressively decreased from CP-A group to CP-C (0.59±0.28 n=6), R2* and FF correlations of MRQuantif to 1.5T Ideal-IQ were r²=0.6 and to 0.25±0.19, p<0.0001) and from MELD≤10 to MELD 19-24 (0.58±0.30 to 0.6, to 3T Vibe-Dixon were r²=0.76 and 0.31 or to 3T mDixon-Quant were 0.54±0.49, p=0.773). SI-HEP had a good performance in distinguishing CP-A r²=0.94 and 0.87, respectively. Correlations increased greatly when excluding from CP-B/C patients (AUC=0.75; 95%CI=0.66-0.83). Among biochemical pa- high liver iron overload with still clinically significant map results errors. 3T R2* rameters, a moderate correlation was found between SI-HEP and total biliru- maps were not accurate above 200 µmol of iron/g liver. bin (R=-0.324), GOT (R=-0318) and albumin (R= 0.320). Comparable results Conclusion: R2* and FF maps are less robust than the ROI analysis proposed were observed when SI-HEP was normalized for liver volume. by MRQuantif.org freeware. Conclusion: SI-HEP of Gd-EOB-DTPA-enhanced MRI studies can be used to effectively evaluate liver function. In clinical practice, this imaging technique could be performed both to correctly characterize liver lesions and to assess the severity of liver disease especially in the perspective of surgical treatment. S692 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 13.7 SS 13.9 MRI hepatic fat quantification using various methods, at Liver surface nodularity quantification from routine CT 1.5 and 3T, in comparison to quantitative images for the detection of clinically significant portal histomorphometry hypertension 1 1 1 1 1 1 2 1 1 A. Paisant , J. Jouan , E. Bardou-Jacquet , E. Bannier , R. Sartoris , P.-E. Rautou , L. Elkrief , V. Vilgrain , M. Ronot ; 2 2 1 1 2 1 2 J. Boursier , C. Aubé , Y. Gandon ; Rennes/FR, Angers/FR Clichy/FR, Geneva/CH Purpose: To compare the results of different methods of quantification by MRI Purpose: To determine whether quantification of liver surface nodularity (LSN), of the hepatic fat fraction (FF), at 1.5T and 3T, with the histological, semi- a computer-based marker derived from routine CT images, can estimate the quantitative and especially quantitative measurement of steatosis. presence of clinically significant portal hypertension (CSPH) in patients with Material and methods: 169 patients were prospectively included in 2 centers, cirrhosis, and to compare LSN with other existing non-invasive tests. one using a 1.5T Signa (GEMS), the other using a 3T Verio (Siemens) MRI. A 2D Material and methods: This retrospective study included 189 patients with gradient echo multi-echo sequence (2D-ME-GRE), designed for hepatic fat cirrhosis who underwent hepatic venous pressure gradient (HVPG) and ab- and iron assessment, was used to quantify FF. MRquantif software calculated dominal CT, including 102 (54%) with CSPH. LSN was measured on CT and Dixon method (dual-echo, triple-echo or with T2* correction) and „complex“ other available non-invasive tests were performed, namely liver and spleen method taking into account a 3-peak spectral modeling of the fat. All patients volumes, liver-to-spleen volume ratio, platelet/spleen diameter, Iranmanesh had a liver biopsy with semi-quantitative assessment and a quantitative meas- score, aspartate aminotransferase/platelet ratio index (APRI), Fibrosis-4 (FIB- urement of FF by determining the percentage of surface of the fat vacuoles. 4). Technical success rate and measurement time were calculated. Area under Analyses used were Spearman test and Bland-Altman comparison. receiver operating characteristic (AUROC) and DeLong test were used. Results Results: MRI and quantitative FF were well correlated: Dixon dual-echo were validated in an independent external cohort of 78 patients with biopsy- (ρs=0.95, p<0.001 at 1.5T; ρs=0.87, p<0.001 at 3T), triple-echo (ρs=0.95, proven cirrhosis. p<0.001 at 1.5T; ρs=0.93, p<0.001 at 3T), corrected by T2* (ρs=0.94, p<0.001 Results: Technical success rate of LSN was 93%; measurement time was at 1.5T ; ρs=0.94, p<0.001 at 3T), complex (ρs=0.94, p<0.001 at 1.5T; ρs=0.94, 113±36.3 seconds. LSN quantification was correlated to the HVPG (r=0.75; p<0.001 at 3T). The biases calculated by the Bland-Altman method were p<.001). Patients with CSPH had a significantly higher LSN score than those closed to 0. Correlation of semi-quantitative histological assessment to MRI without (3.22±0.59 vs. 2.44±0.34; p<.001). A cut-off value of 2.76 had a posi- FF and also to quantitative histological FF was weaker. tive predictive value of 88% for CSPH. AUROC of LSN for the estimation of Conclusion: Using a 2D-ME-GRE sequence, MRQuantif freeware provides a CSPH was 0.88. This was significantly higher than that of other available non- good evaluation of FF compared to a really quantitative gold standard at 1.5T invasive tests (all p<.001). Cut-off value 2.76 was validated in the independent and 3T. MRI FF assessment is a better gold standard than the semi-quantita- external cohort. tive histological grading. Conclusion: Non-invasive CT-based LSN quantification rapidly and accurate - ly detects CSPH. This strategy could be particularly useful to evaluate portal hypertension before liver surgery. SS 13.8 Liver MR relaxometry: basic rules for clinical application and reference values when analyzing T1 and T2* maps SS 13.10 V.C. Obmann, N. Mertineit, C. Marx, A. Berzigotti, Modern MRI in the diagnostic workup of cystic fibrosis- J. Heverhagen, A. Christe, A.T. Huber; Bern/CH associated liver disease S. Pötter-Lang, K. Staufer, P. Baltzer, D. Tamandl, Purpose: To determine T1 mapping reference values of the liver at 3T with N. Bastati-Huber, D. Muin, L. Kazemi-Shirazi, respect of hepatic fat, fibrosis and iron content, as well as susceptibility-in - A. Ba-Ssalamah; Vienna/AT duced liver effects from the adjacent lung parenchyma. Material and methods: 143 patients without solid liver lesions >1cm on con- Purpose: To identify independent imaging features and establish a diagnostic trast-enhanced abdominal CT scans underwent multiparametric MR imaging algorithm for early diagnosis of cystic fibrosis-associated liver disease (CFLD) on a Siemens Verio 3T magnet including T1-/T2*-mapping, proton density fat in CF patients as compared with a control group (CG) on gadoxetic acid-en- fraction (PDFF) quantification and MR elastography. The population was di - hanced MRI. vided into negative patients without fibrosis (shear modulus <2.8 kPa) and Material and methods: A total of 90 adult patients were enrolled, 50 with CF steatosis (PDFF<10%) and positive patients with fibrosis and/or steatosis. T1 and 40 of the CG. The CF group was divided into two subgroups, the retro- values were compared between liver segments and groups. Multivariate analy- spective or test subgroup (n = 33) and the prospective or validation subgroup sis was used to statistically correct T1 for T2*. Mann-Whitney U test between (n = 17). The CG (patients with normal liver enzymes, but no more than benign segments and groups was performed. focal liver lesions) was divided also accordingly (27:13). MRI variables includ- Results: In negative patients, T2*-time was a significant confounder for T1- ing quantitative and qualitative parameters were used to distinguish those with time (p<1^15), while age and sex (p=0.722, 0.687) were not. T1 and T2 relaxa- CFLD from the CG using the clinical criteria by Debray et al. Fifteen qualitative tion times were shorter in lung-adjacent liver segments as compared with non- single-lesion CF descriptors were defined. Two readers independently evalu - lung-adjacent segments (mean of differences 33 ms, p<0.001 for T1). Normal ated the images. Univariate statistical analysis was performed to obtain the T1 in reference patients’ non-lung-adjacent segments was 780±83 ms. In significant imaging features that differentiate CF patients from the CG. Subse - positive patients with steatosis and fibrosis, T1 was significantly higher quently, a multivariate classification analysis using the chi-squared automatic (p=0.011 or less). When correcting T1 for T2*, this discrimination between interaction detector (CHAID) methodology was performed to identify the most positive and negative patients could be improved (p<0.001 for all compari- important descriptors. Diagnostic performance was assessed by receiver-op- sons). erating characteristic (ROC) analysis. Conclusion: When analyzing T1 maps in the liver at 3T, non-lung-adjacent Results: Three independent imaging descriptors distinguished CFLD from CG: segments should be measured due to susceptibility effects of the adjacent a) the presence of altered gallbladder morphology (GBAM); b) periportal track- lung. Liver iron content as indicated by T2*-shortening is a major confounder ing (PPT) and c) periportal fat deposition (PPFD). Prospective validation of the of T1 time. classification algorithm demonstrated a sensitivity of 94.1% and specificity of 84.6% for discriminating CFLD from the control cohort. Conclusion: We identified 3 independent imaging features that could poten - tially diagnose early-stage CFLD on unenhanced MRI. S693 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 Wicklow Hall 2A SS 14.4 Virtual unenhanced imaging of the liver derived from Scientific Session SS 14 160mm rapid-switching dual-energy CT: comparison of Technical advances in abdominal imaging image quality and solid liver lesion conspicuity with true unenhanced images 1 1 1 1 2 M. Lacroix , S. Mulé , E. Hérin , F. Pigneur , P. Richard , SS 14.1 1 1 1 1 B. Zegai , L. Baranes , M. Djabbari , F. Brunetti , withdrawn by the authors 1 1 1 1 N. De Angelis , A. Laurent , A. Rahmouni , A. Luciani ; 1 2 Creteil/FR, Buc/FR SS 14.2 Purpose: To compare virtual unenhanced images derived from arterial (VUEart) Diagnostic accuracy of whole body and portal-venous phases (VUEport) with native unenhanced (NU) acquisitions 18F-fluorodeoxyglucose positron emission tomography- in patients with suspected solid liver tumors (SLT) imaged on a rapid-switching MR in HCCs: comparison of local and metastatic single-source 160mm dual-energy CT (rsDECT). staging with chest-abdomen-pelvic CT and liver MRI Material and methods: 73 patients with 52 SLT who underwent a multiphasic 1 1 1 1 M. Vermersch , B. Emsen , S. Mulé , J. Chalaye , abdominal rsDECT were included in this institutional review board approved 2 1 1 1 1 A. Monnet , E. Hérin , F. Pigneur , L. Baranes , H. Regnault , retrospective study. Both NU (120kVp, 40mm, Pitch 1.375, rotation 0.7s, 1 1 1 1 1 G. Amaddeo , A. Laurent , A. Rahmouni , E. Itti , A. Luciani ; ASIRv 40%), VUEart and VUE port (80 kVp-140 kVp; 80mm collimation, pitch 1 2 Creteil/FR, Saint Denis/FR 0.992, rotation time 0.8s, ASIRv 50%) were reconstructed from raw data. Both Purpose: To assess the diagnostic accuracy and therapeutic management qualitative detectability of SLT on a 4-point scale and quantitative analyses of impact of whole body 18F-fluorodeoxyglucose positron emission tomogra - liver and erector-spinae muscles mean+SD densities, and lesion-to-liver con- phy-MRI (WB-18FDG-PET-MR) for local and whole body staging of HCCs. trast-to-noise-ratio (CNR) were compared on NU, VUEart and VUEport Material and methods: 36 consecutive patients followed for HCC were in- (Kruskal–Wallis, Dunn, Kolmogorov–Smirnov). cluded in this retrospective study. All patients underwent WB-18FDG-PET-MR Results: The mean liver density was significantly lower on VUEart and VUE - (Biograph mMR) which included whole body sequences (simultaneous multi- port than on NU (respectively, 56.7±6.7 and 57.9±6 vs 59.6±7.5; p<0.01), but slice with blipped Controlled Aliasing in Parallel Imaging Results in Higher Ac- was similar on VUEart and VUEport (p=0.09). The mean muscle density was celeration sequence design (SMS diffusion-weighted imaging (DWI)), unen- significantly lower on VUEart and VUEport than on NU (respectively, 47.1±5.4 hanced and post contrast 3DT1 Dixon VIBE) simultaneously acquired with PET and 48.4±5.3 vs 51.8±5.2; p<0.0001), but was similar on VUEart and VUEport data (25 minutes acquisition overall), and dedicated liver sequences (SMS in- (p=0.5). The lesion-to-liver CNR was significantly higher on VUEart and VUE - travoxel incoherent motion (IVIM) DWI, T2, multiphasic 3DT1 VIBE). The ac- port than on NU (respectively, 1.7±1 and 1.6±1.1 vs 0.9±0.6; p<0.001), but was curacy of WB-18FDG-PET-MR for HCC detection, local—biliary, vascular in- similar on VUEart and VUEport (p>0.9). Lesion conspicuity was significantly volvement—regional N staging, and distant staging was compared to that higher on VUEport images than on NU (p=0.004). observed with chest-abdomen-pelvis (CAP) CT and liver MRI, with pathology, Conclusion: rsDECT-derived VUE improves lesion-to-liver CNR and image follow-up or consensus interpretation of all available data as the reference quality and could replace true unenhanced acquisitions in this oncology set- standard. ting. Results: Overall, 55 HCC and 7 metastatic sites were documented in 33 pa- tients. Sensitivity, specificity, positive predictive value (PPV) and negative pre - SS 14.5 dictive value (NPV) of WB-18FDG-PET-MR were 100% for liver HCC detection, Accuracy of multi-parametric MRI in the diagnosis and local and regional staging. The sensitivity of WB-18FDG-PET-MR was 100%, grading of intestinal acute graft-versus-host disease specificity 97%, PPV 83%, NPV 100% and accuracy 97% for detection of after allogenic bone marrow transplantation metastatic patients. WB-18FDG-PET-MR led to changes in therapeutic man- D. Vitale, F. Maccioni, V. Buonocore, D. Bencardino, agement in 14% of patients (5/36), 2 by detection of additional HCC and 3 by M. Lopez, A.P. Iori, C. Catalano; Rome/IT detection of additional metastatic sites. Purpose: The diagnosis of intestinal acute graft-versus-host disease (a- Conclusion: One-stop-shop WB-18FDG-PET-MR improves local and distant GVHD) after allogenic bone marrow transplantation is based on clinical symp- HCC tumor staging compared to CAP CT and liver MRI performed separately, toms, CT findings and biopsies. We assessed MRI diagnostic accuracy. with significant impact on patient treatment management. Material and methods: We performed a prospective study on 35 patients (range 9-69 years) with clinical suspicion of intestinal a-GvHD, who underwent SS 14.3 clinical-endoscopic, histological and multi-parametric MRI evaluation. Inclu- withdrawn by the authors sion criteria: intestinal symptoms within 100 days from transplantation, <2 weeks between MRI and clinical-endoscopic evaluation. The following MRI features were evaluated: small and large bowel wall thickening, T2 signal and gadolinium enhancement of the intestinal wall, mesenteric lymph nodes, peri- toneal effusion, comb-sign and restricted diffusion. Histology, clinical-endo- scopic data and follow-up were considered as gold standard for diagnosis and staging. Sensibility, specificity, accuracy and Cohen’s kappa were calculated. Results: In 21/35 (60%), a-GvHD was confirmed at histology and follow-up. In true-positive patients, MRI showed significant continuous wall thickening in 76.2%, stratified wall contrast enhancement in 90%, comb-sign in 76%, in - creased number of mesenteric lymph nodes in 19%, and free intra-peritoneal fluid in 57.2%. A significant correlation was found between the number of pathological intestinal segments and clinical grade of a-GvHD (r=0.54, p=0.009). The distal ileum was the segment most frequently involved (85.7%). MRI sensitivity was 90.5%, specificity 78%, the PPV 86%, NPV 84% and ac - curacy 85%. Furthermore, MRI detected early disease in 82% of patients with- out intestinal symptoms. Conclusion: In patients with intestinal a-GvHD, MRI can diagnose and grade the disease with high accuracy, in both early and more advanced or severe stages. S694 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 14.6 SS 14.8 How low can we go? Assessment of the lowest CT Transient arterial phase respiratory motion-related enterography dose achievable with iterative artifact in liver MR imaging with gadoxetic acid vs. reconstruction extracellular gadolinium: a within-patient cohort 1 1 2 3 M. Twomey , H. Javed , D. Mondal , S. Moghe , comparative study 1 1 1 1 J. Sammon , P. Rogalla , L. Guimaraes ; Toronto, ON/CA, E. Belmonte, A. Darnell, J. Rimola, A. Forner, V. Sapena, 2 3 Oxford/UK, Vernon Hills, IL/US J. Bruix, C. Ayuso; Barcelona/ES Purpose: To determine the maximum dose reduction achievable without sac- Purpose: To compare the frequency and severity of transient arterial phase rificing observer performance for the detection of Crohn’s disease (CD) lesions (TAP) respiratory motion-related artifact following gadoxetic acid (EOB) and in CT enterography (CTE) using filtered back projection (FBP), hybrid iterative extracellular gadolinium (Ec-Gd) in a cohort of cirrhotic patients. reconstruction (H-IR) and a new model-based IR (MB-IR) algorithm. Material and methods: Eighty-two cirrhotic patients from prospective series Material and methods: 50 CTE studies performed in a 320 row scanner for underwent liver MR with EC-Gd and EOB within 1 month. Two readers, blinded known/suspicion of CD between July 2016 and July 2017 were included. A to all data, assigned a respiratory motion-related sore (0 [none], 1 [mild, no validated noise insertion tool was utilized to simulate 75%, 50% and 25% of impact on reader confidence interpretation], 2 [moderate, causing decrease in original dose. Datasets were reconstructed with FBP, H-IR and 2 types of MB- reader confidence interpretation], and 3 [sever, non-diagnostic]) for non-en - IR (IR1 and IR2). Two fellowship-trained abdominal radiologists independently hanced, arterial and portal venous phase. Scores for each phase were aver- reported intestinal findings and assessed image quality (IQ), noise and sharp - aged, and definitions for TAP, severe TAP and post-contrast motion-related ness using 5-point scales. Assessment of the full-dose dataset/endoscopic artifacts were established. The frequency of motion-related artifacts was com- data by an experienced abdominal radiologist served as reference standard. pared for each pair of examinations. Results: Fifty patients with 48 CD lesions were included. Mean effective dose Results: The mean motion scores for EC-Gd and EOB for non-enhanced of the full-dose datasets was 3.9 mSv. At 50% & 75% dose, sensitivity was phase were 0.68 vs 0.73 (p=0.3), for arterial phase 0.91 vs. 1.26 (p<0.0001), 98% for both readers and all reconstruction methods. At 25% dose sensitivi- and for portal venous phase 0.73 vs. 0.87 (p=0.04). The frequency of TAP ob- ties of FBP, H-IR, MB-IR were 88%, 92%, 88% for Reader 1 and 84%, 96%, served on EOB was superior to EC-Gd (19.5% vs. 6%; p=0.013), but there 89% for Reader 2. Agreement was almost perfect (k=0.82). IQ of FBP was were no differences in frequencies for severe TAP (p>0.99) between both con- unacceptable and worse than H-IR and MB-IR (p<0.05) at all dose levels. trasts agents. EOB was associated with a higher frequency of post-contrast Mean H-IR and MB-IR IQ scores were <3 (acceptable) at 50% and 75% dose. motion-related artifacts (7.3% vs. 0%; p=0.03). No significant difference was seen between H-IR and MB-IR1/2 (p>0.1357). Conclusion: TAP respiratory motion-related artifact is more frequently ob- MB-IR1 images were the sharpest, while MB-IR2 were less noisy (p<0.05). served on liver MRI with EOB than with EC-Gd and it may affect 19.5% of the Conclusion: Low-dose and ultralow-dose CTE can be performed with H-IR patients. However, the rate of severe TAP was very low and similar for both and MB-IR with acceptable image quality without compromising observer per- contrast agents. formance, providing knowledge facilitating confident implementation of lower dose CTE protocols using multiple CT platforms and reducing radiation expo- SS 14.9 sure in this patient population. withdrawn by the authors SS 14.7 SS 14.10 The effect of gadoxetic acid on image quality of multiple Assessment of whole-body fat using MRI: a pilot study hepatic arterial phase imaging: comparison study with 1 2 2 2 J.-P. Kühn , N. Friedrich , M.-L. Kromrey , J. Effler , gadoteric acid-enhanced MRI 1 1 3 1 R.-T. Hoffmann , M. Laniado , R. Laqua ; Dresden/DE, K.C. Sim, B.J. Park, H. Na Yeon, M.J. Kim, D.J. Sung, 2 3 Greifswald/DE, Bern/CH S.B. Cho; Seoul/KR Purpose: To develop an algorithm to assess fat compartments of the whole Purpose: To evaluate the effect of gadoxetic acid on image degradation of body using confounder-corrected chemical shift-encoded magnetic reso- single-breath-hold, triple (first, second, and third) arterial phase acquisition of nance imaging (CSI-MRI). the liver compared with those obtained with gadoteric acid. Material and methods: Seventy-three volunteers aged 29 to 80 years (29 fe- Material and methods: A total of 211 consecutive MR examinations were male) underwent a whole-body MRI including three-echo chemical shift-en- evaluated including triple arterial phase acquisition using 4D eTHRIVE Keyhole coded sequences acquired in 5 steps. After confounder correction and stitch- technique. 108 examinations were performed with gadoxetic acid and 103 ex- ing of steps, the whole-body proton-density fat fraction was acquired. Based aminations were performed with gadoteric acid. Two radiologists evaluated on tissue fat content, we developed a post-processing for segmentation of fat image qualities of the precontrast and triple arterial phases of both groups in a compartments as well as for tissue fat quantification. Using this algorithm, two blinded fashion. Image quality on each phase was assessed in regard to image observers measured total body fat volume (TAT), subcutaneous fat volume artifacts, sharpness of intrahepatic vessel, sharpness of liver edge, and overall (SAT), visceral fat volume (VAT), and liver fat content (LFC). Robustness of the image quality. algorithm was tested by evaluation of inter- and intraobserver variability using Results: Gadoxetic acid group shows better image score in the precontrast a Bland-Altman analysis. In addition, data were correlated with body mass in- images, despite the compromised conditions such as cirrhosis, more ascites, dex (BMI) and age using Spearmen correlation coefficients. and variable comorbid malignancies that could make more degraded image Results: TAT, SAT, FAT and LFC were successfully measured in all volunteers. quality (p < 0.05). Image quality of triple arterial phases of gadoxetic acid-en- There were no differences between two readings (mean bias (%)±standard hanced liver MRI was significantly degraded than those of gadoteric acid for deviation; TAT/SAT/FAT/LFC: -0.1±0.3/0.4±1.0/0.2±0.4/0.4±1.0) and both ob- artifact, sharpness of intrahepatic vessel, sharpness of liver edge, and overall servers (TAT/SAT/FAT/LFC: 0.0±0.3/0.9±1.4/0.4±0.8/0.9±1.6). TAT showed the quality in each arterial phases (p < 0.05). Overall image quality was gradually highest correlation with BMI (r=0.88, p<0.01), followed by SAT (r=0.83, p<0.01), improved from first to third arterial phase (p < 0.003). Third arterial phase VAT (r=0.76, p<0.01), and LFC (r=0.61, p<0.01). The order was not affected by showed best image quality in both groups. Inter-rater reliability was high (p < sex; however, the correlation coefficients were generally higher in men than in 0.001). women. Conclusion: Intravenous gadoxetic acid can have a detrimental effect on im- Conclusion: CSI-MRI is an excellent approach to assess whole-body fat vol- age quality of triple arterial phases MR imaging using 4D eTHRIVE keyhole ume and organ fat content. It opens new perspectives for clinical research in technique. Third arterial phase images show best image quality regardless of imaging of metabolic disorders. the contrast media. S695 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 Wicklow Hall 2B SS 15.2 Diagnostic performance of dual-tracer positron Scientific Session SS 15 emission tomography-CT with 18F-fluorodeoxyglucose Liver and bile duct intervention: evaluating and 11C-acetate versus modified response evaluation criteria in solid tumors in the assessment of residual treatment response viable tumours in HCC that received loco-regional therapy SS 15.1 K.W.H. Chiu, T.-T. Cheung, R. Lo, H. Yuan, L.W. Chan; Preoperative CT findings for prediction of incomplete Hong Kong/HK resection and poor survival in patients with gallbladder Purpose: Dual-tracer positron emission tomography (DT-PET) with 11C-ace- cancer 1 2 1 1 1 tate (Ac) and 18F-fluorodeoxyglucose (FDG) have shown to be highly sensitive J.H. Kim , S.-Y. Choi , H.J. Park , H.W. Eun , J.K. Han ; 1 2 in the detection of HCC although thus far, it has only been applied to treat- Seoul/KR, Bucheon/KR ment-naïve patients. The aim of this study was to evaluate the diagnostic per- Purpose: To predict residual tumor (R) classification and poor survival in pa - formance DT-PET in assessing HCC that had undergone loco-regional treat- tients who underwent surgery for gallbladder cancer using preoperative CT. ment. Material and methods: From 2006 to 2012, 173 patients (M:F=96:77, Material and methods: Patients with HCC who received loco-regional treat- 65.3years) with gallbladder cancer underwent CT and surgery. Two radiolo- ment were retrospectively identified and included if they underwent CT or MRI gists assessed CT findings including tumor morphology, location, T-stage, ad - and DT-PET within 3 months of resection of the lesion. All lesions were evalu- jacent organ invasion, hepatic artery abutment, portal vein encasement, LN ated on cross-sectional imaging using the modified response evaluation crite - metastasis, metastasis, resectability, gallstone, and combined cholecystitis. R ria in solid tumours (mRECIST) and radiotracer uptake on PET were recorded classification was categorized into no residual tumor (R0) and residual tumor and compared with histology. (R1 or R2). We analyzed the correlation between CT findings and R classifica - Results: Fourteen patients with 30 HCC fulfilled the inclusion criteria; 11 un - tion. In addition, we followed up the patients up to 5 years and analyzed the derwent transarterial chemoembolization (TACE), 2 radiofrequency ablation relationship between each CT findings and overall survival using Cox regres - (RFA) and 1 both RFA and TACE (median time from previous intervention 3 sion analysis and Kaplan–Meier analysis. months, range 1-43 months). Residual tumour was present in 24 lesions on Results: R classification included no residual tumor in 127 patients and re - histology. Using mRECIST criteria, 16/30 lesions had complete response, 14 sidual tumor in 46 patients. 5-year survival was 44%. For prediction of R clas- partial response/stable disease. FDG was avid in 15/30, Ac in 20/30 and DT sification, hepatic invasion (Exp(B)=4.008, 95% CI, 1.22-6.39, p=0.014) and PET in 22/30 lesions. The positive predictive values for both modalities were hepatic artery abutment (Exp(B)= 8.67, 95% CI, 3.24-25.21, p=0.001) were 100% and the sensitivities in detecting residual disease were 54.2% and significant predictors for residual tumor. On survival analysis, residual tumor 83.3% (p<0.05) for cross-sectional imaging and DT-PET, respectively. (Exp(B)= 2.68, 95% CI, 1.72-4.19, p=0.001), liver metastasis (Exp(B)= 6.18, Conclusion: mRECIST criteria on cross-sectional imaging under-detect re- 95% CI, 1.41-27.07, p=0.016), duodenum invasion (Exp(B)= 2.8, 95% CI, 1.61- sidual viable tumour in HCC that had undergone loco-regional treatment. DT- 4.87, p=0.001), and regional LN metastasis (Exp(B)= 6.18, 95% CI, 1.41-27.07, PET is a sensitive modality that can be used as an alternative modality for the p=0.016) were significant predictors for poor survival. treatment reassessment especially in cases where patients’ management may Conclusion: Preoperative CT findings could aid planning surgery using the be altered. high-risk findings of residual tumor including hepatic invasion and hepatic ar - tery abutment. In addition, CT can predict poor survival using important find - SS 15.3 ings including liver metastasis, duodenum invasion, and regional LN metasta- CT texture analyses can predict early regrowth after sis. thermal ablation of colorectal liver metastases D.J. Van Der Reijd, E. Klompenhouwer, M. Taghavirazavizadeh, F. Staal, F. Imani, D. Meek, M.C. De Boer, D.M.J. Lambregts, R.G.H. Beets-Tan, M. Maas; Amsterdam/NL Purpose: To assess whether CT texture analysis can detect early regrowth of colorectal liver metastases (CRLM) one month after thermal ablation. Material and methods: 38 patients with CRLM who underwent radiofrequen- cy ablation (RFA) or microwave ablation (MWA) for lesions ≤3 cm were includ- ed. Ablation zones 1-month post-ablation were manually delineated on all slices on portal venous phase CT by an expert radiologist blinded to clinical outcome. A margin of 5 mm was mathematically added to the ablation zone and texture parameters (entropy, uniformity and mean grey-level intensity) were compared between patients with and without regrowth using multiple Laplacian of Gaussian filters ( σ0.5-2.5). Analyses were performed with pyradi- omics. Results: 18/38 patients had a regrowth. Mean follow-up was 32 (standard deviation (SD) 27) months and mean time to regrowth was 9 (SD 8) months. Entropy of the ablation zone (AZ) and ablation margin (AM) were significantly lower in regrowths (mean 1.34 vs 1.71, p=0.004 and mean 1.47 vs. 1.81, p=0.044, respectively). Uniformity of both the AZ and AM were higher in re- growths (mean 0.45 vs. 0.37, p=0.004 and mean 0.42 vs. 0.35, p=0.03, respec- tively). These differences were found both for analyses with a filter of σ0.5 and without a filter. For mean grey-level intensity no significant differences were observed for neither the AZ nor AM (mean 1.22 vs 0.73, p=0.496 and -1.90 vs -2.04, p=0.874, respectively). Conclusion: Patients with regrowth after ablation show a lower entropy and higher uniformity of the ablation zone and ablation margin 1-month post-abla- tion. CT texture analyses can help in identifying patients who are at risk for regrowth. S696 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 15.4 SS 15.7 Influence of pre-treatment tumor growth rate on the Factors associated with tumor progression after objective response of HCC treated with selective percutaneous ablation of HCC: results of a propensity trans-arterial chemoembolization score matching analysis comparing monopolar Y. Purcell, R. Pommier, V. Roche, M. Dioguardi Burgio, radiofrequency and microwaves A. Sibert, V. Vilgrain, M. Ronot; Clichy/FR D. Bouda, V. Barrau, C. Garcia-Alba, V. Roche, M. Dioguardi Burgio, A. Sibert, V. Vilgrain, M. Ronot; Purpose: To assess the influence of pre-treatment tumor growth rate ( TGR) pre Clichy/FR on the objective response (OR) according to modified response evaluation cri - teria in solid tumors (mRECIST) criteria after a first session of selective trans- Purpose: To identify risk factors for local and distant tumor progression after arterial chemoembolization (TACE) for the treatment of HCC. percutaneous thermal ablation of HCC with a focus on the comparison be- Material and methods: Ninety-nine patients (91 male, 93%) with 148 tumors tween microwave ablation (MWA) and monopolar radiofrequency ablation (mean 34±29mm), who underwent the first session of selective TACE for the (RFA). treatment of HCC between 2011 and 2016, were included. preTGR was calcu- Material and methods: Data were reviewed from consecutive cirrhotic pa- lated as the percentage change in tumour volume per month (%/mo) on avail- tients with very early/early HCCs who underwent image-guided percutaneous able CTs before treatment. OR was defined as a complete or partial response monopolar expandable RFA or MWA. Factors associated with local and distant on a lesion-based analysis. preTGR cut-off for prediction of OR was identified tumor progression were identified using propensity score matching (PSM). by receiver operating characteristic (ROC) curve analysis. Statistical analysis was performed with the Kaplan-Meier method using the Results: After one session of treatment 74 (50%), and 41 (28%) tumors log-rank test and Cox regression models. showed complete (CR), or partial response (PR) while 29 (20%) and 4 (2%) Results: A total of 190 HCC (mean 23±8.6mm) in 149 patients (mean 63±11 showed stable (SD), or progressive disease (PD). The mean TGR was 13±17 y-o, 79% male) underwent RFA (n=90, 47%) or MWA (n=100, 53%). Second- pre %/mo. It followed a U-shaped distribution with a mean 15±19, 13±16, 6±10, ary technical success rate was 97.4%. After a median follow-up of 24.6 and 15±5 %/mo in tumors showing CR, PR, SD, and PD. After exclusion of months (interquartile range (IQR): 9.7-37.2), 43 (23%) HCC showed local pro- tumors showing PD, TGR was significantly higher in tumors with OR gression (median time delay 13.4 months (IQR: 5.8-24.3)) and 91 (63%) pa- pre (mean14±18 %/mo) when compared to those showing SD (6±10 %/mo, tients had a distant progression (median time delay 10.4 months (IQR: 5.7-22). p=0.03). A TGR value of 6.5%/mo was associated with a sensitivity and Cox model after PSM identified treatment by RFA (hazard ratio (HR), 2.934; pre specificity of 60% and 86% for the prediction of OR (Area under ROC of P=0.006), size ≥30mm (HR, 3.130; P=0.007) and vascular contact (HR, 3.372; 0.67±0.05). P=0.005) as risk factors for local progression. Factors associated with distant Conclusion: Pre-treatment TGR is highly variable in HCC before TACE with a progression were a main tumor size ≥30mm (HR, 1.94; P=0.013), an alpha-fe- U-shaped distribution for the prediction of tumor response. It provides insight toprotein (AFP) level above 100 ng/mL (HR, 2.56; p=0.002), and hepatitis B into tumor biology that may be used during pre-treatment workup to help (HR, 0.51; p=0.047). stratify patients. Conclusion: The rate of local HCC progression was lower after MWA when compared to monopolar expandable RFA, regardless of tumor size and vascu- lar contact. Ablation technique did not influence the risk of distant progression. SS 15.5 withdrawn by the authors SS 15.8 Predicting the response of colorectal cancer liver SS 15.6 metastases to preoperative chemotherapy using Radiological tumor response and histopathological gadoxetic acid-enhanced MRI correlation of HCC treated with stereotactic body 1 2 2 2 N.V. Costa , N. Bastati , S. Pötter-Lang , Z. Guengoern , radiation therapy as a bridge to liver transplantation 2 2 1 2 Y. Bican , A. Ba-Ssalamah ; Lisbon/PT, Vienna/AT K. Mastrocostas, S. Fischer, P. Munoz-Schuffenegger, Purpose: The aim of the study was to create a scoring system based upon H.-J. Jang, G. Sapisochin, L.A. Dawson, T.K. Kim; ga gadoxetic acid-enhanced MRI ( MRI) features to predict the treatment re- Toronto, ON/CA sponse (TR) to chemotherapy. Purpose: To assess the imaging appearance of HCC treated with stereotactic Material and methods: This was a retrospective study of 30 consecutive pa- body radiation therapy (SBRT) as a bridging therapy prior to liver transplanta- tients (65.2±11.2years) with colorectal cancer liver metastases (CRCLM), who tion (LT), with pathological correlation. underwent gaMRI after chemotherapy and before hepatic resection. Metasta- Material and methods: 23 patients (20 males; mean age, 60 years) were in- ses were classified according to a suggested scoring system (0-6 points) in cluded, with a total of 25 lesions assessed. CT and/or MRI imaging pre- and three groups of response: optimal (≤2 points); partial (2-4 points); and no-re- post-SBRT were reviewed by two independent radiologists, with a third reader sponse (≥4 points). The scoring system comprised three features: overall ho- resolving inter-observer differences. The target lesions were assessed for ra- mogeneity (homogeneous=0, mixed=1, heterogeneous=2); tumor liver inter- diological response including change in size, change in attenuation in the por- face (sharp=0, mixed=1, ill-defined=2); and peripheral rim enhancement (≤2 tal venous phase on CT or signal intensity on MRI, and percentage necrosis if mm=0, 2-4 mm=1, ≥4 mm=2). Apparent diffusion coefficient (ADC) values were present. The radiological findings were compared to the tumor necrosis at measured. The primary outcome was residual vital tumor (RVT). The scoring explant as assessed by a pathologist. system, response groups, and ADC values were calculated and compared Results: Of the 25 lesions, 15 lesions were classed as stable disease (SD), 8 with the RVT percentage. Demographic, laboratory, and imaging findings were lesions as partial response (PR) and 2 lesions as progressive disease (PD). included in a multivariate statistical analysis. The three groups of response Among the 15 SD lesions, all lesions were identified as having necrosis at ex - were correlated with patient survival and the log-rank test was used to com- plant pathology and imaging. Ten of these lesions were assessed as having pare two survival distributions (optimal/partial responses vs no-response imaging changes of necrosis concordant within 10% of the necrosis at pathol- groups). ogy. Nine lesions were assessed at imaging as having >20% necrosis, all of Results: Forty-one CRCLM showed good inter-observer agreement (κ=0.86). which were concordant with the pathology findings, despite stable size. In only Multiple regression demonstrated an association between RVT (32.9±11.2) one case was necrosis at imaging >20% above that at explant pathology. and the scoring system (p<0.001), the response group (p<0.001), and the ADC Conclusion: Current standard HCC response criteria such as response evalu- values (p<0.021). The survival distributions between optimal/partial response ation criteria in solid tumors (RECIST) underestimate the response of lesions to and no-response showed a trend to be different (p=0.066). local therapy with SBRT. As the utility of SBRT as a focal therapy for HCC in- ga Conclusion: MRI correlated well with our scoring system, different response creases, further evaluation of the imaging findings of response is required. groups, and ADC values in patients with CRCLM treated with chemotherapy, and may be used to assess the RVT percentage. S697 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 15.9 SS 15.10 Early prediction of response to transarterial Early recurrence of HCC predicted by whole-lesion chemoembolization using CT perfusion in primary liver radiomics on preoperative gadoxetic acid-enhanced tumors MRI D. Tamandl, F. Waneck, A. Ba-Ssalamah, C. Loewe; Z. Zhang, J. Chen, H. Jiang, L. Cao, T. Duan, Z. Ye, B. Song; Vienna/AT Chengdu/CN Purpose: Assessment of CT perfusion (CTP) in the prediction of response Purpose: To investigate the role of gadoxetic acid (Gd-EOB-DTPA)-enhanced early after transarterial chemoembolization (TACE) in patients with HCC and MRI-derived texture features in predicting early recurrence (≤6 months) of cholangiocarcinoma (CCC). HCC. Material and methods: This prospective study was approved by the institu- Material and methods: 29 consecutive patients with preoperative Gd-EOB- tional review board and written informed consent was obtained from all pa- DTPA-enhanced MRI were included. Early recurrence was determined within 6 tients. We performed CTP one day before and after (interquartile range [IQR]: months. 3D texture features on individual sequence were derived using an in- 1-2 days) TACE in 18 patients. 46 lesions (41 HCC and 5 hypervascular CCC) house software (Analysis Kit, GE Healthcare). LASSO logistic regression mod- were amenable for analysis. Various perfusion-related parameters were meas- el was adopted for feature selection. Identified texture features were compared ured. In a lesion-based analysis, absolute and relative changes were com- between patients with and without early recurrence. Receiver-operating char- pared to the clinical response on biphasic CT using modified response evalu - acteristic (ROC) analysis of predicting early recurrence was performed for each ation criteria in solid tumors (mRECIST) criteria 6 weeks after the procedure. texture feature. Optimal cutoff values for detecting response were calculated using area under Results: 17 early recurrences were confirmed with imaging follow-up. The the ROC curves (AUROC). Follow-up analysis included assessment of overall skewness, sumEntropy, ShortRunHighGreyLevelEmphasis_AllDirection_off- survival (OS), local recurrence-free survival (LRFS) and local progression-free set7_SD, and LongRunLowGreyLevelEmphasis_AllDirection_offset7_SD of survival (LPFS), using Kaplan-Meier estimates. HCC on arterial-phase, the ShortRunHighGreyLevelEmphasis_AllDirection_ Results: The clinical response was: complete response (CR, n=22, 48%), par- offset4_SD and LongRunHighGreyLevelEmphasis_angle90_offset7 on portal tial response (PR, n=9, 20%) and no response (14 stable, 1 progression) 6 venous-phase (PVP), and uniformity and LongRunLowGreyLevelEmphasis_ weeks after TACE. CTP parameters were all reduced after TACE in responding AllDirection_offset7_SD on hepatobiliary phase (HBP) were found to be statis- patients (PR, CR) while no difference was observed in non-responders. Arte- tically different between patients with and without recurrence (P < 0.05). The rial liver perfusion (ALP ) was superior in prediction of CR compared to blood areas under the ROC curves (AUCs) were 0.966 (CI[0.82, 0.99]), 0.725 (CI[0.53, post flow (BF ) and blood volume (BV ) (AUROC 0.953 vs. AUROC 0.859 0.87]), 0.760 (CI[0.57, 0.89]) and 0.833 (CI[0.65, 0.95]) for the skewness, su- post post ALP BF and AUROC 0.831, p<0.001) with a sensitivity, specificity, positive predictive mEntropy, ShortRunHighGreyLevelEmphasis_AllDirection_offset7_SD, and BV value, negative predictive value and accuracy of 91%, 92%, 91%, 92% and LongRunLowGreyLevelEmphasis_AllDirection_offset7_SD on AP, 0.789 91%. Only 4/22 lesions with CR recurred with a median local recurrence-free (CI[0.59, 0.92]) and 0.814 (CI[0.63, 0.93]) for the ShortRunHighGreyLevelEm- survival of 22.7 months. phasis_AllDirection_offset4_SD and LongRunHighGreyLevelEmphasis_an- Conclusion: Early response assessment after transarterial chemoemboliza- gle90_offset7 on PVP, and 0.843 (CI[0.66, 0.99]) and 0.760 (CI[0.63, 0.93]) for tion can reliably be performed with CT perfusion. the uniformity and LongRunLowGreyLevelEmphasis_AllDirection_offset7_SD on HBP, respectively, in predicting early recurrence of HCC. The highest sensi- tivity and specificity were shown on arterial-phase images (skewness) of 91.7% (CI[61.5%,99.8%]) and 94.1% (CI[71.3%,99.9%]). Conclusion: Whole-lesion texture features based on Gd-EOB-DTPA-en- hanced-MRI held promise for predicting early recurrence of HCC after hepa- tectomy, especially the skewness on arterial-phase images. S698 Scientific Sessions Authors’ Index S699 Scientific Sessions Authors’ Index AUTHORS’ INDEX Beets-Tan R.G.H.: SS 2.4, SS 2.6, SS 4.3, Cao L.: SS 1.9, SS 9.3, SS 15.10 SS 4.10, SS 5.2, SS 9.1, SS 9.4, Caramia E.: SS 4.5 SS 9.9, SS 15.3 Cardano G.: SS 3.9, SS 8.7 Aalbers A.G.J.: SS 4.3 Behari J.: SS 7.2 Cardobi N.: SS 4.7, SS 8.4, SS 11.2 Aerts H.J.W.L.: SS 9.1, SS 9.4 Bektas C.T.: SS 5.9 Carrozzo V.: SS 5.1 Albazaz R.: SS 1.1, SS 13.3 Bellini D.: SS 5.8, SS 6.1, SS 12.10 Cartier V.: SS 7.1, SS 7.7 Albrecht M.H.: SS 4.8 Belmonte E.: SS 14.8 Caruso D.: SS 6.1 Aleman B.M.P.: SS 5.2 Bencardino D.: SS 14.5 Carvalho L.: SS 11.1 Alexander D.: SS 2.3 Benea G.: SS 12.9 Casiraghi A.: SS 6.10 Alfaro I.: SS 6.5 Berbee M.: SS 9.9 Cassinotto C.: SS 7.7 Amaddeo G.: SS 14.2 Bergamasco L.: SS 4.5, SS 5.1, SS 10.7 Castera L.: SS 7.5 Ambrosetti M.C.: SS 11.3 Bertuzzo L.: SS 3.9, SS 8.7, SS 8.10, Castro J.: SS 6.5 Andersson M.: SS 8.3 SS 11.8, SS 11.10 Catalano C.: SS 14.5 Anysz-Grodzicka A.: SS 9.5 Berzigotti A.: SS 13.8 Cazals-Hatem D.: SS 12.6 Aslan S.: SS 8.1 Beuzit L.: SS 7.3 Cecchin D.: SS 2.7 Atkinson D.: SS 6.2 Beykoz Çetin E.: SS 7.10 Chalaye J.: SS 14.2 Atri M.: SS 9.2 Bican Y.: SS 15.8 Chan L.W.: SS 15.2 Aubé C.: SS 7.1, SS 7.7, SS 13.7 Biscaldi E.: SS 12.4 Chang W.: SS 8.2 Auger M.: SS 7.3 Bissoli A.: SS 11.3 Chargeboeuf C.: SS 13.6 Axelsson E.: SS 8.3 Boddenberg J.: SS 11.9 Chen J.: SS 9.3, SS 10.4, SS 15.10 Ayuso C.: SS 1.5, SS 14.8 Bodelle B.: SS 4.8 Chermak F.: SS 7.7 Boellaard T.N.: SS 2.6 Cheung T.-T.: SS 15.2 Bogner P.: SS 6.7 Chiu K.W.H.: SS 15.2 Ba-Ssalamah A.: SS 5.4, SS 7.9, SS 13.5, Bonatti G.: SS 10.9 Cho S.B.: SS 14.7 SS 13.10, SS 15.8, SS 15.9 Bonatti M.: SS 10.9 Choi M.H.: SS 5.5 Badia S.: SS 5.8 Borg P.: SS 11.1 Choi S.-Y.: SS 1.8, SS 9.8, SS 15.1 Bae J.S.: SS 8.2, SS 9.8 Borhani A.: SS 7.2, SS 10.10 Chowdry P.: SS 5.10 Bakers F.C.H.: SS 2.4, SS 4.10, SS 9.1, Bouda D.: SS 15.7 Christe A.: SS 13.8 SS 9.4, SS 9.9 Boudiaf M.: SS 6.6 Chryssou E.: SS 1.10 Baldwin-Cleland R.: SS 12.1 Boulay-Coletta I.: SS 3.4, SS 3.8 Churilov L.: SS 5.10 Bali M.A.: SS 5.8 Bouquot M.: SS 6.6 Ciaravino V.: SS 4.7, SS 11.2 Ball C.: SS 12.7 Boursier J.: SS 7.1, SS 7.7, SS 13.7 Cieszanowski A.: SS 9.5 Baltzer P.: SS 13.10 Bowden D.: SS 3.7 Cingarlini S.: SS 11.2 Bannier E.: SS 13.7 Bowden D.J.: SS 3.2 Clarke C.G.D.: SS 1.1, SS 12.2 Baranes L.: SS 14.2, SS 14.4 Bozzato A.M.: SS 4.5 Çolakoğlu M.K.: SS 7.10 Barat M.: SS 3.5, SS 6.6 Bradley K.: SS 8.6 Collins C.: SS 5.3 Barata M.J.: SS 2.5, SS 2.8 Brown P.J.: SS 2.1, SS 4.9 Conci S.: SS 4.7 Barbieri S.: SS 12.9 Bruix J.: SS 1.5, SS 14.8 Contro A.: SS 4.7 Bardou-Jacquet E.: SS 13.6, SS 13.7 Brun V.: SS 7.3 Conwell D.: SS 11.7 Barra F.: SS 12.4 Brunetti F.: SS 14.4 Corno L.: SS 3.4, SS 3.8 Barrau V.: SS 15.7 Bruno O.: SS 1.7 Corr A.: SS 2.9 Bartels-Rutten A.: SS 5.2 Brusic A.: SS 5.10 Costa F.: SS 11.1 Bartholomä W.: SS 8.3 Bruzzi J.: SS 5.3 Costa N.V.: SS 15.8 Bassi C.: SS 8.9 Buonocore V.: SS 14.5 Cournane S.: SS 4.2 Bassi M.: SS 12.9 Burling D.: SS 2.9, SS 12.1 Cressoni M.: SS 4.1 Bastati N.: SS 15.8 Byrne D.: SS 6.8 Crimì F.: SS 2.7 Bastati-Huber N.: SS 7.9, SS 13.5, Byrne J.: SS 3.3 Crona J.: SS 11.1 SS 13.10 Cronin C.: SS 6.8 Bazeries P.: SS 7.1 Cronin C.G.: SS 4.4 Beable R.: SS 12.7 Cui J.: SS 9.10 Calame P.: SS 3.5 Beer A.: SS 7.9 Cybulski A.J.: SS 3.9 Cales P.: SS 7.7 Beer L.B.: SS 13.5 Camlidag İ.: SS 8.1 Beets G.L.: SS 2.6, SS 9.4 Cannella R.: SS 7.2, SS 10.10 Cantwell C.P.: SS 3.10 S700 Scientific Sessions Authors’ Index AUTHORS’ INDEX Grimm R.: SS 8.4 D F Grossarth V.: SS 11.9 Guengoern Z.: SS 15.8 D’Onofrio M.: SS 4.7, SS 8.4, SS 8.9, Fabiszewska E.: SS 9.5 Guimaraes L.: SS 14.6 SS 11.2 Fabris C.: SS 8.10, SS 11.3, SS 11.10 Guler E.: SS 3.1 Darnell A.: SS 1.5, SS 14.8 Faccinetto A.: SS 7.5 Gursoy Coruh A.: SS 2.10 Darwish O.: SS 7.6 Fahlenkamp U.: SS 10.5 Guthrie J.A.: SS 1.1, SS 13.3 Dasyam A.: SS 11.7 Faki A.A.: SS 3.6 Dautry R.: SS 6.6 Faletti R.: SS 4.5, SS 5.1, SS 10.7 David M.: SS 12.6 Faluhelyi N.: SS 6.7 Dawson L.A.: SS 15.6 Famularo S.: SS 13.4 Haimerl M.: SS 13.2 De Angelis N.: SS 14.4 Farkas O.: SS 6.7 Halligan S.: SS 12.1 De Arcos J.: SS 7.6 Feeney J.: SS 12.3 Hamilton G.: SS 9.10 De Boer M.C.: SS 15.3 Fehrenbach U.: SS 10.5 Hammerstingl R.: SS 4.8 De Ledinghen V.: SS 7.7 Feier D.S.: SS 7.9 Han J.K.: SS 8.2, SS 9.8, SS 15.1 De Mestier L.: SS 11.1 Fenlon H.: SS 4.4, SS 6.8 Hanbidge A.: SS 3.3 De Paoli Barbato G.: SS 12.9 Ferraris A.: SS 4.5 Harguem S.: SS 10.8 De Robertis R.: SS 4.7, SS 8.4, SS 8.9, Ferrero S.: SS 12.4 Harman M.: SS 3.1 SS 11.2 Fidler J.L.: SS 8.6 Hatzidakis A.A.: SS 1.10, SS 4.6 De Santis D.: SS 6.1 Fighera A.: SS 3.9 Haug A.: SS 5.4 Del Chiaro M.: SS 8.3 Figueiredo N.: SS 2.5, SS 2.8 Healy G.M.: SS 3.10 Delahaye J.: SS 7.1 Fiore S.: SS 4.5 Heidous M.: SS 3.6 Deledda A.: SS 12.9 Fischer S.: SS 15.6 Henderson W.C.: SS 9.10 Delli Pizzi A.: SS 2.6 Fletcher J.G.: SS 8.6 Hérin E.: SS 14.2, SS 14.4 Denecke T.: SS 10.5 Fogel E.L.: SS 11.5, SS 11.7 Heverhagen J.: SS 13.8 Dijkhoff R.A.P.: SS 2.4, SS 4.10 Fonio P.: SS 4.5, SS 5.1, SS 10.7 Higginson A.: SS 12.7 Dioguardi Burgio M.: SS 7.5, SS 15.4, Forner A.: SS 1.5, SS 14.8 Hill A.: SS 3.2, SS 3.7 SS 15.7 Fouchard I.: SS 7.7 Hiriart J.-B.: SS 7.7 Djabbari M.: SS 14.4 Franca R.: SS 11.1 Hoeffel C.: SS 6.6 Doblas S.: SS 12.6 Frantsev D.: SS 10.6 Hoffmann R.-T.: SS 11.6, SS 14.10 Dohan A.: SS 6.6 Friedrich N.: SS 11.6, SS 14.10 Hooker J.C.: SS 9.10 Dolgushin B.: SS 10.6 Frigerio I.: SS 8.9 Houssel-Debry P.: SS 7.3 Drago S.G.: SS 2.4, SS 4.1, SS 4.10, Fritz F.: SS 8.5 Huber A.T.: SS 13.8 SS 6.10 Frostberg E.: SS 12.8 Hupkens B.J.P.: SS 2.6 Dromain C.: SS 11.1 Fukuda H.: SS 10.3 Hwang J.A.: SS 1.8 Duan T.: SS 1.9, SS 10.4, SS 15.10 Furlan A.: SS 7.2, SS 10.10 Dubois M.: SS 7.3 Füger B.: SS 5.4 Dunne R.: SS 3.2, SS 3.7 Đuri ć-Stefanović A.: SS 8.8 Ianus A.: SS 2.2, SS 2.3 Ibukuro K.: SS 10.3 Gandon Y.: SS 7.3, SS 13.6, SS 13.7 Imani F.: SS 15.3 Garcia-Alba C.: SS 15.7 Imbault M.: SS 7.5 Effler J.: SS 14.10 García-Criado A.: SS 1.5 Iori A.P.: SS 14.5 Einspieler H.: SS 7.9 Garg I.: SS 8.6 Ippolito D.: SS 4.1, SS 6.10, SS 13.4 Elkrief L.: SS 13.9 Garteiser P.: SS 12.6 Itti E.: SS 14.2 Elmas N.Z.: SS 3.1 Gatti M.: SS 4.5, SS 5.1, SS 10.7 Elmer M.: SS 13.5 Geisel D.: SS 10.5 Emsen B.: SS 14.2 Gennisson J.-L.: SS 7.5 Engbersen M.: SS 4.3 Gianotti L.: SS 13.4 Jain R.: SS 7.4 Erden A.: SS 2.10 Gibney R.: SS 1.3, SS 10.2 Jang H.-J.: SS 9.2, SS 13.1, SS 15.6 Ernst O.: SS 3.8 Giganti M.: SS 12.9 Jasieniak J.: SS 9.5 Esler S.: SS 5.10 Girelli R.: SS 8.9 Javed H.: SS 14.6 Eun H.W.: SS 15.1 Goenka A.: SS 8.6 Jenkins J.T.: SS 2.9 Eveno C.: SS 6.6 Gołębiewski B.: SS 9.5 Jeon C.Y.: SS 11.5 Gollifer R.: SS 6.2 Jeon S.J.: SS 9.8 Greer P.: SS 11.7 Jeong W.K.: SS 1.4, SS 10.1 S701 Scientific Sessions Authors’ Index AUTHORS’ INDEX Jevtić S.: SS 8.8 Laghi A.: SS 5.8, SS 6.1, SS 12.10 Mansueto G.: SS 3.9, SS 8.7, SS 8.10, SS 11.3, SS 11.8, SS 11.10 Jiang H.: SS 9.3, SS 15.10 Lahaye M.J.: SS 4.3, SS 9.1, SS 9.4, SS 9.9 Marchegiani G.: SS 3.9, SS 8.10 Johnson G.B.: SS 8.6 Johnson S.: SS 9.2 Lamarca A.: SS 11.1 Maretto I.: SS 2.7 Lambregts D.M.J.: SS 2.4, SS 2.6, Martel A.: SS 9.2 Johnston C.: SS 4.2 SS 4.3, SS 4.10, SS 5.2, SS 9.1, Joo I.: SS 8.2 Martin S.S.: SS 4.8 SS 9.4, SS 9.9, SS 15.3 Marx C.: SS 13.8 Jouan J.: SS 13.6, SS 13.7 Landoni L.: SS 11.2 Jung S.E.: SS 5.5 Mastrocostas K.: SS 15.6 Laniado M.: SS 11.6, SS 14.10 Masulovic D.: SS 8.8 K Lannes A.: SS 7.7 Matos C.: SS 2.2, SS 2.3, SS 2.5, SS 2.8, Lapteva M.: SS 10.6 SS 9.7 Kaci R.: SS 6.6 Lapuyade B.: SS 7.7 Mayer P.F.J.: SS 8.5 Kadıoğlu M.E.: SS 7.10 Laqua R.: SS 14.10 McCann J.W.: SS 3.10 Kalarakis G.: SS 1.10, SS 4.6 Lauenstein T.: SS 11.9 McCormick P.A.: SS 1.3, SS 10.2 Kalcan S.: SS 7.10 Laurent A.: SS 14.2, SS 14.4 McLoughlin L.: SS 4.2 Kale S.: SS 7.4 Lawlor R.: SS 9.2 McQuade C.S.: SS 4.4 Karantanas A.: SS 1.10, SS 4.6 Lazić L.: SS 8.8 Meek D.: SS 15.3 Kartalis N.: SS 8.3 Le Bail B.: SS 7.7 Melisi D.: SS 8.9 Kauczor H.U.: SS 8.5 Lebert P.: SS 3.8 Menys A.: SS 6.2 Kavanagh R.G.: SS 5.6 Lebigot J.: SS 7.1, SS 7.7 Mertineit N.: SS 13.8 Kazemi-Shirazi L.: SS 13.10 Lee A.: SS 3.2, SS 3.7 Metin Y.: SS 7.10 Keane R.: SS 1.3 Lee D.H.: SS 5.7, SS 7.8 Metric Investigators T.: SS 6.9 Keussen I.: SS 8.3 Lee J.Y.: SS 7.8 Michalak S.: SS 7.7 Khalili K.: SS 9.2 Lee M.: SS 3.2, SS 3.7, SS 12.5 Middleton M.S.: SS 9.10 Kiani A.: SS 3.5 Lee S.: SS 10.1 Miles A.: SS 6.9 Kilickesmez O.: SS 5.9 Lee S.M.: SS 5.7 Millet I.: SS 3.4, SS 3.8 Kim J.E.: SS 6.4, SS 9.8 Leithner D.: SS 4.8 Milliat F.: SS 12.6 Kim J.H.: SS 8.2, SS 9.8, SS 15.1 Lenga L.: SS 4.8 Min J.H.: SS 1.4 Kim K.W.: SS 10.1 Li L.: SS 11.5 Minervini M.I.: SS 10.10 Kim M.J.: SS 14.7 Li M.: SS 11.4 Moghe S.: SS 14.6 Kim S.H.: SS 5.7, SS 6.4 Liddy S.: SS 12.3 Moloney F.: SS 5.6 Kim S.S.: SS 1.8 Lin X.: SS 10.4 Mondal D.: SS 14.6 Kim T.K.: SS 9.2, SS 13.1, SS 15.6 Lo R.: SS 15.2 Monnet A.: SS 14.2 Kim T.O.: SS 6.4 Loewe C.: SS 15.9 Montesano M.: SS 12.10 Kim Y.K.: SS 1.4 Loizou L.: SS 8.3 Monti M.L.: SS 11.8 Kishino M.: SS 10.3 Lombardi S.: SS 6.10 Morrin M.: SS 3.2, SS 3.7, SS 12.5 Klauss M.: SS 8.5 Lombardo F.: SS 10.9 Mouries A.: SS 7.7 Klompenhouwer E.: SS 15.3 Lopez C.: SS 11.1 Muin D.: SS 13.10 Kocak B.: SS 5.9 Lopez M.: SS 14.5 Mulé S.: SS 14.2, SS 14.4 Kok N.F.M.: SS 4.3 Lourenço J.M.G.: SS 9.7 Mulsow J.: SS 4.4 Kokkinos N.: SS 4.6 Luciani A.: SS 14.2, SS 14.4 Munoz-Schuffenegger P.: SS 15.6 Kose T.: SS 3.1 Lundell L.: SS 8.3 Murphy A.N.: SS 4.4 Kosidekakis N.: SS 4.6 Murray T.É.: SS 12.5 Kovac J.: SS 8.8 Kromrey M.-L.: SS 11.6, SS 14.10 M B.: SS 7.4 Kühn J.-P.: SS 11.6, SS 14.10 Maas M.: SS 2.4, SS 2.6, SS 4.10, Na Yeon H.: SS 14.7 Kukołowicz P.: SS 9.5 SS 5.2, SS 9.1, SS 9.4, SS 9.9, Najran P.S.: SS 11.1 Kumar D.: SS 3.6 SS 15.3 Nederveen A.J.: SS 7.6 Kundaragi N.: SS 7.4 Maccioni F.: SS 14.5 Negrelli R.: SS 8.10 Kuś P.: SS 9.5 Maher M.: SS 5.6 Nikolić T.: SS 8.8 Maher P.: SS 5.10 L Nolan N.: SS 1.3, SS 10.2 Maheux A.: SS 10.8 Nural M.S.: SS 8.1 Lacognata C.: SS 2.7 Malone D.E.: SS 1.3, SS 10.2 Lacroix M.: SS 14.4 Manhoopi Y.: SS 12.8 Lagadec M.: SS 3.5 S702 Scientific Sessions Authors’ Index AUTHORS’ INDEX Plumb A.: SS 6.2, SS 12.1 O S Pocard M.: SS 6.6 Podgorska J.: SS 9.5 O Brien C.M.: SS 10.2 Saint-Jalmes H.: SS 13.6 Poelsterl S.: SS 13.2 O’Brien A.C.: SS 1.3, SS 3.10, SS 10.2 Salvia R.: SS 8.9 Polcaro A.: SS 7.6 O’Brien C.: SS 1.3 Sammon J.: SS 3.3, SS 14.6 Pomerri F.: SS 2.7 O’Brien S.: SS 5.6 Samreen N.: SS 8.6 Pommier R.: SS 15.4 O’Connor O.J.: SS 5.6 Sankar M.: SS 11.5 Pötter-Lang S.: SS 7.9, SS 13.5, O’Dwyer E.: SS 12.3 Santiago I.: SS 2.2, SS 2.3, SS 2.5, SS 13.10, SS 15.8 O’Leary P.: SS 3.7 SS 2.8, SS 9.7 Pourafkari M.: SS 9.2 Santinha J.: SS 2.2, SS 4.10, SS 9.7 O’Malley E.: SS 5.3 Pozzi Mucelli R.: SS 8.7, SS 11.8, Sapena V.: SS 14.8 O’Neill A.: SS 12.3 SS 11.10 O’Rourke C.: SS 4.2 Sapisochin G.: SS 15.6 Prasad V.: SS 10.5 Sarkar R.: SS 12.3 O’Shea A.: SS 12.5 Purcell Y.: SS 15.4 O’Suilleabhain C.: SS 5.6 Sarno A.: SS 4.7, SS 8.9, SS 11.2 Sartoris R.: SS 13.9 Obaro A.E.: SS 12.1 Oberti F.: SS 7.7 Savoldi E.: SS 12.7 Scala C.: SS 12.4 Obmann V.C.: SS 13.8 Quehen E.: SS 7.3 Scarpa A.: SS 11.2 Ogasawara G.: SS 10.3 Ogier-Denis E.: SS 12.6 Schaefer N.: SS 11.1 Schneider T.: SS 7.6 Okrainec A.: SS 3.3 Rafaelsen S.R.: SS 12.8 Opalinska M.: SS 11.1 Schurink N.: SS 9.9 Rahmouni A.: SS 14.2, SS 14.4 Segersvärd R.: SS 8.3 Ordas I.: SS 6.3 Orhan Metin N.: SS 7.10 Rahr H.: SS 12.8 Seltman T.A.: SS 11.5 Raimondi E.: SS 12.9 Sergeeva O.: SS 10.6 Orsini E.B.: SS 13.4 Ramanathan S.: SS 3.6 Shah Z.: SS 11.7 Özdemir O.: SS 7.10 Ramotar H.: SS 13.3 Ozutemiz O.: SS 3.1 Sheehan M.: SS 3.2, SS 3.7 Rangaswamy B.: SS 10.10 Sheehy N.: SS 4.2 P Rautou P.-E.: SS 1.7, SS 7.5, SS 13.9 Shemesh N.: SS 2.2, SS 2.3 Rayar M.: SS 7.3 Sheridan M.B.: SS 1.1, SS 13.3 Paiella S.: SS 8.9 Reddy N.: SS 7.4 Shields C.: SS 4.4 Paisant A.: SS 7.3, SS 13.6, SS 13.7 Regnault H.: SS 14.2 Shili S.: SS 7.7 Pałucki J.: SS 9.5 Rengo M.: SS 5.8, SS 12.10 Shin H.C.: SS 1.8 Panchal N.: SS 7.4 Reynolds J.V.: SS 4.2 Shorikov M.A.: SS 10.6 Panés J.: SS 6.3, SS 6.5 Ricart E.: SS 6.3, SS 6.5 Shur J.: SS 2.9 Papadakis A.: SS 1.10 Richard P.: SS 14.4 Sibert A.: SS 15.4, SS 15.7 Papanikolaou N.: SS 2.2, SS 4.10, SS 9.7 Rimola J.: SS 1.5, SS 6.3, SS 6.5, Sim K.C.: SS 14.7 Parés O.: SS 2.5, SS 2.8 SS 14.8 Sinkus R.: SS 7.6 Park B.J.: SS 14.7 Rivosecchi F.: SS 6.1 Sirlin C.: SS 9.10 Park H.J.: SS 15.1 Rizzati R.: SS 12.9 Sironi S.: SS 4.1, SS 6.10 Park S.J.H.: SS 9.8 Roche V.: SS 15.4, SS 15.7 Skehan S.J.: SS 1.3, SS 10.2 Pasicz K.: SS 9.5 Rodriguez S.: SS 6.3 Skornitzke S.: SS 8.5 Patil A.: SS 3.10 Roe C.: SS 12.2 Skrzyński W.: SS 9.5 Pautrat K.: SS 6.6 Rogalla P.: SS 14.6 Sleeman M.: SS 5.10 Pavel M.: SS 10.5, SS 11.1 Rondenet C.: SS 3.4 Smith C.R.: SS 1.1, SS 13.3 Pecorelli A.: SS 4.1, SS 13.4 Ronot M.: SS 1.7, SS 3.5, SS 7.5, Song B.: SS 1.9, SS 9.3, SS 10.4, Pedersen M.R.: SS 12.8 SS 10.8, SS 11.1, SS 13.9, SS 15.4, SS 15.10 Peker E.: SS 2.10 SS 15.7 Soyer P.: SS 6.6 Perin A.: SS 2.7 Rossington H.: SS 2.1, SS 4.9 Staal F.: SS 15.3 Perisinakis K.: SS 1.10, SS 4.6 Rousset S.: SS 10.7 Staufer K.: SS 13.10 Persohn S.A.: SS 11.5 Rowe I.: SS 1.1 Stemmer A.: SS 8.4 Pezzutti D.: SS 11.1 Runge J.H.: SS 7.6 Stiller W.: SS 8.5 Picchia S.: SS 5.8, SS 12.10 Rutledge N.: SS 3.10 Stirling A.D.: SS 4.4 Pierce B.: SS 1.3 Ruzzenente A.: SS 4.7 Stoker J.: SS 6.2 Pigneur F.: SS 14.2, SS 14.4 Ryan R.: SS 1.3, SS 10.2 Stoyanova D.P.: SS 13.5 Plessier A.: SS 1.7 Strinnholm J.: SS 8.3 S703 Scientific Sessions Authors’ Index AUTHORS’ INDEX Stroszczynski C.: SS 13.2 Van’t Sant I.: SS 4.3 Zerunian M.: SS 6.1 Sundin A.: SS 11.1 Vas D.: SS 6.5 Zhang Z.: SS 15.10 Sung D.J.: SS 14.7 Vasin D.V.: SS 8.8 Ziayee N.: SS 11.9 Szeverenyi N.: SS 9.10 Vellone V.: SS 12.4 Zins M.: SS 3.4, SS 3.8 Verfürth F.: SS 11.9 Zucchetta P.: SS 2.7 Vermersch M.: SS 14.2 Vidal Trueba H.: SS 11.1 Taghavirazavizadeh M.: SS 15.3 Vietti Violi N.: SS 11.1 Takahashi N.: SS 11.7 Vilana R.: SS 1.5 Talei Franzesi C.: SS 4.1, SS 6.10, Vilgrain V.: SS 1.7, SS 3.5, SS 7.5, SS 13.4 SS 10.8, SS 12.6, SS 13.9, SS 15.4, Talwade R.: SS 7.4 SS 15.7 Tamandl D.: SS 5.4, SS 13.5, SS 13.10, Vipperla K.: SS 11.7 SS 15.9 Virshke E.: SS 10.6 Tang G.: SS 11.7 Vitale D.: SS 14.5 Tanter M.: SS 7.5 Vliegen R.F.A.: SS 9.1 Taourel P.: SS 3.4, SS 3.8 Vogl T.J.: SS 4.8 Taylor S.A.: SS 6.2, SS 6.9 Vollenbrock S.E.: SS 5.2 Tedesco G.: SS 4.7, SS 8.9, SS 11.2 Voncken F.E.M.: SS 5.2 Ter Beek L.C.: SS 5.2 Vos F.: SS 6.2 Territo P.R.: SS 11.5 Thornton E.: SS 12.5 Tilli M.: SS 12.9 Tinazzi Martini P.: SS 8.4, SS 11.2 Wald R.M.: SS 13.1 Tirkes T.: SS 11.5, SS 11.7 Walker S.P.: SS 13.3 Tobe K.: SS 10.3 Walsh J.P.: SS 6.8 Tolan D.J.M.: SS 2.1, SS 4.9, SS 12.2 Waneck F.: SS 15.9 Topazian M.: SS 11.7 Wang M.: SS 1.9 Treanor D.: SS 1.1 Waters C.: SS 5.3 Trebeschi S.: SS 9.4 Welaratne I.: SS 4.2 Troelstra M.A.: SS 7.6 Wichmann J.L.: SS 4.8 Truty M.: SS 8.6 Wieszczy P.: SS 9.5 Tsung A.: SS 10.10 Wiggermann P.: SS 13.2 Tublin M.: SS 7.2 Wilson D.: SS 13.3 Turkcanoglu M.H.: SS 5.9 Wyatt J.I.: SS 1.1 Turlin B.: SS 7.3 Twomey M.: SS 5.6, SS 14.6 Yadav D.: SS 11.7 Yang H.K.: SS 13.1 Ugarte-Cano C.: SS 12.1 Yang N.: SS 5.10 Yardımcı A.H.: SS 5.9 Ye Z.: SS 1.6, SS 15.10 Yoo S.-J.: SS 13.1 Valletta R.: SS 10.9 Yoon J.-H.: SS 6.4 Valls C.: SS 8.3 Yu H.: SS 11.4 Van Beers B.: SS 7.5, SS 12.6 Yuan H.: SS 15.2 Van Der Reijd D.J.: SS 15.3 Van Der Sande M.: SS 2.6 Van Dieren J.M.: SS 5.2 Van Eden H.: SS 4.3 Zagórowicz E.: SS 9.5 Van Elmpt W.J.C.: SS 9.9 Zamboni G.A.: SS 3.9, SS 8.7, SS 8.10, Van Griethuysen J.: SS 2.4, SS 2.6, SS 10.9, SS 11.3, SS 11.8, SS 11.10 SS 4.10, SS 9.9 Zanirato M.: SS 8.4 Van Griethuysen J.J.M.: SS 9.1, SS 9.4 Zappa M.: SS 3.5, SS 12.6 Van Wettere M.: SS 1.7 Zegai B.: SS 14.4 S704 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Insights into Imaging Springer Journals

ESGAR 2018 Book of Abstracts

Insights into Imaging , Volume 9 (2) – May 29, 2018
Free
54 pages
Loading next page...
 
/lp/springer_journal/esgar-2018-book-of-abstracts-QXZrWSvJAb
Publisher
Springer Berlin Heidelberg
Copyright
Copyright © 2018 by The Author(s)
Subject
Medicine & Public Health; Imaging / Radiology; Diagnostic Radiology; Interventional Radiology; Neuroradiology; Ultrasound; Internal Medicine
eISSN
1869-4101
D.O.I.
10.1007/s13244-018-0634-1
Publisher site
See Article on Publisher Site

Abstract

Insights into Imaging Education and strategies in European radiology ESGAR 2018 Book of Abstracts / Volume 9 / Supplement 2 / June 2018 ESGAR 2018 / June 12 – 15 / Dublin, Ireland th 29 Annual Meeting and Postgraduate Course t th h 2 29 9 A AN NN NU UA AL M L ME EE ET TIIN NG G A AN ND P D PO OS ST TG GR RA AD DU UA AT TE C E CO OU UR RS SE E BOOK OF ABSTRACTS INCLUDES ABSTRACTS OF SCIENTIFIC PRESENTATIONS J J J J J J J JU U U U U U U UN N N N N N N NE 1 E 1 E 1 E 1 E 1 E 1 E 1 E 12 2 2 2 2 2 2 2 - - - - - - - - 1 1 1 1 1 1 1 15 5 5 5 5 5 5 5 Eur European opean E ES SG GA AR R 2 20 01 18 8 Society Society of Gastr of Gastrointestinal ointestinal D D D D D D D DU U U U U U U UB B B B B B B BL L L L L L L LI I I I I I I IN N N N N N N N and Abdominal and Abdominal IIIIIIIIIIR R R R R R R R R RE E E E E E E E E EL L L L L L L L L LA A A A A A A A A AN N N N N N N N N ND D D D D D D D D D Radiology Radiology IMPORTANT ADDRESSES / CME / CONTRIBUTING SOCIETIES / CORPORATE MEMBERS ORGANISING SECRETARIAT CONFERENCE VENUE Central ESGAR Office The Convention Centre Dublin Esslinggasse 2/3 Spencer Dock AT – 1010 Vienna North Wall Quay Phone: +43 1 535 89 27 IE - D01 T1W6 Dublin 1 Fax: +43 1 535 89 27 -15 E-Mail: office@esgar.org WEBSITE www.esgar.org CME CREDITS th The “ESGAR 2018 29 Annual Meeting and Postgraduate The “ESGAR European Society of Gastrointestinal and Course, Dublin, Ireland, 12/06/2018-15/06/2018” has th Abdominal Radiology” (or) “ESGAR 2018 – 29 ANNUAL been accredited by the European Accreditation Council for MEETING AND POSTGRADUATE COURSE” is accredited Continuing Medical Education (EACCME®) with 25 by the European Accreditation Council for Continuing European CME credits (ECMEC®s). Each medical Medical Education (EACCME) to provide the following specialist should claim only those hours of credit that he/ CME activity for medical specialists. The EACCME is an she actually spent in the educational activity. institution of the European Union of Medical Specialists (UEMS), www.uems.net. CONTRIBUTING SOCIETIES SPONSORS ESGAR wishes to gratefully acknowledge the support of its Corporate Members: The Final Programme of ESGAR 2018 is available on the ESGAR Website www.esgar.org Date of publishing: June 2018 S653 $/)ªS  8-0634-1 TABLE OF CONTENTS TABLE OF CONTENTS Scientific Sessions, Wednesday, June 13 (SS 1 – SS 5) S658-S671 Scientific Sessions, Thursday, June 14 (SS 6 – SS 10) S672-S685 Scientific Sessions, Friday, June 15 (SS 11 – SS 15) S686-S698 Authors’ Index S699-S704 S654 654 COMMITTEES ESGAR 2018 MEETING PRESIDENT ESGAR EXECUTIVE COMMITTEE PRESIDENT Prof. Helen Fenlon S. Halligan, London/UK University College Dublin Department of Radiology PRESIDENT-ELECT Mater Misericordiae University Hospital and BreastCheck R.G.H. Beets-Tan, Amsterdam/NL 36 Eccles Street IE – Dublin 7 VICE PRESIDENT A. Laghi, Rome/IT ESGAR 2018 PROGRAMME COMMITTEE SECRETARY/ TREASURER M. Zins, Paris/FR CHAIRMEN M. Zins, Paris/FR PAST PRESIDENT C. Matos, Lisbon/PT C. Matos, Lisbon/PT MEMBERS EDUCATION COMMITTEE M.A. Bali, London/UK S. Jackson, Plymouth/UK A. Ba-Ssalamah, Vienna/AT MEMBERSHIP COMMITTEE R.G.H. Beets-Tan, Amsterdam/NL M.A. Bali, London/UK G. Brancatelli, Palermo/IT H. Fenlon, Dublin/IE RESEARCH COMMITTEE S. Halligan, London/UK J. Stoker, Amsterdam/NL S. Jackson, Plymouth/UK A. Laghi, Rome/IT WORKSHOP COMMITTEE G. Brancatelli, Palermo/IT J. Stoker, Amsterdam/NL V. Vilgrain, Clichy/FR MEETING PRESIDENT 2018 C.J. Zech, Basel/CH H. Fenlon, Dublin/IE PRE-MEETING PRESIDENT 2019 ESGAR 2018 LOCAL ORGANISING COMMITTEE A. Laghi, Rome/IT R. Browne, Dublin/IE MEMBERS AT LARGE C. Cronin, Dublin/IE A. Ba-Ssalamah, Vienna/AT T. Geoghegan, Dublin/IE V. Vilgrain, Clichy/FR M. Keogan, Dublin/IE C.J. Zech, Basel/CH M. Lee, Dublin/IE M. Maher, Cork/IE ESGAR EXECUTIVE DIRECTOR D.E. Malone, Dublin/IE B. Lindlbauer, Vienna/AT M. McNicholas, Dublin/IE M. Morrin, Dublin/IE LANGUAGE ABSTRACT EDITOR J. Murphy, Galway/IE C. Clarke, Nottingham/UK A. Ryan, Waterford/IE M. Ryan, Cork/IE T. Scanlon, Limerick/IE N. Sheehy, Dublin/IE M. Shelly, Limerick/IE S.J. Skehan, Dublin/IE M. Staunton, Cork/IE S655 ABSTRACT REVIEWING PANEL O. Akhan, Ankara/TR P. Prassopoulos, Alexandroupolis/GR C. Aubé, Angers/FR E. Quaia, Edinburgh/UK M.A. Bali, London/UK G.A. Rollandi, Genova/IT I. Bargellini, Pisa/IT M. Ronot, Clichy/FR T.V. Bartolotta, Palermo/IT W. Schima, Vienna/AT A. Ba-Ssalamah, Vienna/AT S. Schmidt Kobbe, Lausanne/CH R.G.H. Beets-Tan, Amsterdam/NL A. Schreyer, Regensburg/DE E. Biscaldi, Genova/IT O. Seror, Bondy/FR A. Blachar, Tel Aviv/IL S.J. Skehan, Dublin/IE G. Brancatelli, Palermo/IT M. Staunton, Cork/IE D.J. Breen, Southampton/UK S. Stojanovic, Novi Sad/RS F. Caseiro Alves, Coimbra/PT J. Stoker, Amsterdam/NL N. Courcoutsakis, Alexandroupolis/GR J.P. Tasu, Poitiers/FR L. Crocetti, Pisa/IT S.A. Taylor, London/UK L. Curvo-Semedo, Coimbra/PT S. Terraz, Geneva/CH R. Dondelinger, Liège/BE D.J.M. Tolan, Leeds/UK M. D’Onofrio, Verona/IT C. Triantopoulou, Athens/GR H. Fenlon, Dublin/IE V. Valek, Brno/CZ A. Furlan, Pittsburgh, PA/US V. Vilgrain, Clichy/FR Y. Gandon, Rennes/FR M.-P. Vullierme, Clichy/FR V. Goh, London/UK D. Weishaupt, Zurich/CH S. Gourtsoyianni, London/UK G.A. Zamboni, Verona/IT S. Gryspeerdt, Roeselare/BE C.J. Zech, Basel/CH L. Guimaraes, North York, ON/CA M. Zins, Paris/FR J.A. Guthrie, Leeds/UK S. Halligan, London/UK A. Hatzidakis, Heraklion/GR T. Helmberger, Munich/DE F. Iafrate, Rome/IT M. Karcaaltincaba, Ankara/TR N. Kartalis, Stockholm/SE H.-U. Laasch, Manchester/UK A. Laghi, Rome/IT M. Laniado, Dresden/DE J.M. Lee, Seoul/KR M. Lewin, Paris/FR O. Lucidarme, Paris/FR M. Maas, Amsterdam/NL A. Madureira, Porto/PT M. Maher, Cork/IE D.E. Malone, Dublin/IE T. Mang, Vienna/AT V. Maniatis, Aabenraa/DK D. Marin, Durham, NC/US L. Martí-Bonmatí, Valencia/ES C. Matos, Lisbon/PT Y. Menu, Paris/FR G. Morana, Treviso/IT G.H. Mostbeck, Vienna/AT A. Palkó, Szeged/HU P. Paolantonio, Rome/IT N. Papanikolaou, Lisbon/PT R. Pozzi Mucelli, Verona/IT S656 Scientific Sessions Authors’ Index S657 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 1.3 11:00 - 12:30 The Auditorium Radiological-pathological correlation of pre-transplant imaging and explanted livers: what is the prevalence of Scientific Session SS 1 HCC in patients with liver imaging reporting and data Focal liver lesions: new trends in imaging system 2017 category 5, 4/5 and 4 lesions? of HCC A.C. O‘Brien, R. Keane, B. Pierce, C. O‘Brien, N. Nolan, P.A. McCormick, D.E. Malone, S.J. Skehan, R. Ryan, R. Gibney; Dublin/IE SS 1.1 Comparison of European association for the study of the Purpose: The Irish National Centre for Liver Transplantation (OLT) serves a pop- liver, liver imaging and reporting data system criteria and ulation of 4.5 million. Liver imaging reporting and data system (LI-RADS) catego- radiological opinion for diagnosing HCC in cirrhotic livers ry designation is made at a multidisciplinary meeting by 1 or more of 4 abdominal 1 2 2 2 C.G.D. Clarke , R. Albazaz , C.R. Smith , I. Rowe , radiologists. The aim of this study is to evaluate the prevalence of HCC in OLT 2 2 2 2 D. Treanor , J.I. Wyatt , M.B. Sheridan , J.A. Guthrie ; patients whose pre-OLT imaging showed LI-RADS 4 or 5 lesion(s). 1 2 Nottingham/UK, Leeds/UK Material and methods: OLT patients with a pre-operative diagnosis of HCC and explants with HCC found at pathology from 01/01/2014 to 09/09/2017 were Purpose: Imaging criteria for the non-invasive diagnosis of HCC have been es- identified using the OLT database. Corresponding pre-OLT imaging reports (CT/ tablished by many organisations. The reported diagnostic performance of MRI in MRI) were reviewed. In patients with LI-RADS 2017 class 5, 4/5 or 4 lesions, the the diagnosis of HCC is widely variable. Our aim was to determine the sensitivity histopathology coding of explanted livers was reviewed. Data were collated in an and specificity of these criteria by correlating with a histological diagnosis from MSExcel spreadsheet and analysed on a per patient basis in 2 groups (LI-RADS whole liver explants. 5 and LI-RADS 4/5 or 4). Material and methods: This is a single-centre, retrospective review. Participants Results: 53 OLT patients met selection criteria. Pre-operative imaging reported were selected based on the following: consecutive adults (>=18yrs) listed for LI-RADS 5, 4/5 or 4 lesions in 41 of 53 explanted livers. LI-RADS 5: n = 36 pa- liver transplantation in 2014/2015, with cirrhosis at the time of MR scanning with tients (7 had pre-OLT TACE), 35 had HCC, 1 had macroregenerative cirrhosis hepatocyte-specific contrast agent, and at least one liver lesion >=1cm on MR without HCC; 1 also had a collision tumour (HCC/cholangiocarcinoma). HCC with histology from subsequent liver explant for comparison. Patients were ran- prevalence: 97%. LI-RADS 4/5: n =1 patient, HCC confirmed at OLT. LI-RADS 4: domly allocated to two independent consultant radiologists for review. Each le- n = 4 patients, 3 had HCC and dysplastic nodules only were identified in the sion was assessed against international criteria [European association for the other patient. HCC prevalence 80%. study of the liver (EASL), liver imaging and reporting data system (LIRADS)], and Conclusion: The prevalence of HCC in OLT patients with LI-RADS 2017 class 5 given a ‚radiologist impression‘ score of 1-5 (1=definitely benign, 5=definitely lesions is acceptably high. The small LI-RADS 4/5 or 4 group had a lower HCC HCC). prevalence. More data are needed for optimal management of these patients. Results: Total of 268 patient records were reviewed, with 118 eligible lesions identified from 50 patients. Median lesion size was 15.5 (interquartile range 12- 21) mm. Mean age 56 (+/-7.7) yrs with M:F ratio of 4:1. Sensitivity, specificity and SS 1.4 PPV for EASL was 0.47, 0.82 and 0.83, for LIRADS LR5 was 0.39, 0.88 and 0.85, Prospective intraindividual comparison of gadoxetic acid for LIRADS LR4+5+TIV was 0.73, 0.69, and 0.81, and for ‚radiologist impression‘ and gadoterate meglumine for the diagnosis of HCCs with of probably or definitely HCC, available in 108 lesions, was 0.79, 0.79 and 0.88. liver imaging reporting and data system 1 2 1 1 2 Conclusion: MR imaging has moderate sensitivity and good specificity in the Y.K. Kim , J.H. Min , W.K. Jeong ; Seoul/KR, Daejeon/KR diagnosis of HCC and there is considerable variation depending on the criteria Purpose: This prospective study intraindividually compared the efficacy of ga - used. ‘Radiologist impression’ has better sensitivity when compared to LIRADS doxetic acid (Gd-EOB-DTPA)-enhanced MRI and gadoterate meglumine (Gd- and EASL. DOTA)-enhanced MRI for the diagnosis of HCC with the liver imaging reporting and data system (LI-RADS). SS 1.2 Material and methods: Between November 2016 and November 2017, we pro- spectively included 91 patients with chronic liver disease who underwent both withdrawn by the authors Gd-EOB-DTPA-MRI and Gd-DOTA-MRI (interval range, 9 to 30 days) for first detected hepatic nodule on US: 107 lesions (95 HCCs, 2 cholangiocarcinomas, 7 hemangiomas, 2 dysplastic nodules, 1 neuroendocrine tumor; size range: 0.8- 3.0 cm) were identified by surgical resection. Two observers reviewed two MRIs based on the LI-RADS v2017. Results: We found a tendency toward higher sensitivity (observer 1, 76.8% vs 57.9% and observer 2, 81.1% vs. 61.3%) and accuracy (observer 1, 81.2% vs 65.8% and observer 2, 84.6% vs. 68.7%) with LR-5 category on Gd-DOTA-MRI than with that on Gd-EOB-DTPA-MRI (P < 0.001). When applying arterial hyper- enhancement with either washout on 3-min Gd-EOB-DTPA MRI or isointensity with capsule on 3 min Gd-DOTA-MRI, 16 HCCs on Gd-EOB-DPTA-MRI and 10 HCCs on Gd-DOTA-MRI were upgraded to LR-5 category while achieving 100% specificity with both MRIs. One cholangiocarcinoma was correctly classified with Gd-EOB-DTPA MRI due to targetoid appearance. Conclusion: In the diagnosis of HCC with LI-RADS, MRI using extracellular con- trast agent showed better sensitivity and accuracy than Gd-EOB-DTPA-MRI al- though they showed 100% specificity. We could achieve better diagnostic per - formance with applying washout on 3 min Gd-EOB-DTPA-MRI or isointensity with capsule on 3-min Gd-DOTA MRI than conventional criteria, while maintain- ing 100% specificity. S658 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 1.5 SS 1.7 Prospective evaluation of dynamic MR with gadoxetic Hypervascular tumors in patients with primary Budd- acid for the non-invasive diagnosis of HCC in newly Chiari syndrome: is the washout sign specific for HCC? detected nodules An MRI study C. Ayuso, A. Darnell, J. Rimola, A. García-Criado, R. Vilana, A. M. Van Wettere, O. Bruno, P.-E. Rautou, A. Plessier, Forner, J. Bruix; Barcelona/ES V. Vilgrain, M. Ronot; Clichy/FR Purpose: To evaluate the diagnostic accuracy of gadoxetic acid-MR (EOB-MR) Purpose: To reappraise the imaging characteristics of focal liver lesions (FLL) in for non-invasive HCC diagnosis in liver nodules <2 cm detected by screening patients with Budd-Chiari syndrome (BCS), with a focus on the diagnostic value US. of washout for the differentiation between benign and malignant tumors. Material and methods: Cirrhotic patients with newly detected solitary nodules Material and methods: Between 2000 and 2016, patients with a diagnosis of <2 cm by US were included. After written informed consent, extracellular con- BCS and FLL on MRI were included. MRI was retrospectively reviewed by two trast-MR (EC-MR) and EOB-MR were performed in less than 1 month. Final di- radiologists blinded to the nature of these lesions. Patients and tumor character- agnosis was based on the validated European Association for the Study of the istics were recorded, with a focus on signal intensity on MRI sequences. HCC Liver (EASL)/American Association for the Study of Liver Diseases (AASLD) and benign tumors were compared using Fisher‘s test or a Chi-squared test and guidelines criteria (specific vascular profile on EC-MR or biopsy). Blind, double the Student‘s t test or Mann-Whitney test. reading was performed in EOB-MR studies. Criteria for HCC diagnosis using Results: 49 patients (34 female, 69%) with 241 benign lesions and 12 HCC were EOB-MR were arterial contrast uptake and either washout in the portal phase or analyzed. Patients with HCC were significantly older (mean 44±16 vs. 33±9, p hypointensity in the hepatobiliary phase (HBp). =.005), with higher baseline serum alpha-fetoprotein (AFP) level (median 16 [2- Results: 62 consecutive cirrhotic patients Child-Pugh A (53) or B (9) were in- 9000] vs. 3 [2-25] ng/mL, p =.007). HCC was significantly larger (mean 32±16 vs. cluded. Final diagnosis was: 41 (66.1%) HCC, 2 intrahepatic cholangiocarcino- 11±5mm, p<0.001), more frequently hypointense on T1-weighted (58% vs. 4%, ma (ICC), 1 colo-rectal cancer metastasis, and 18 benign conditions. Patients p=.001) and hyperintense on T2-weighted images (58% vs. 18%, p=.003) than with benign lesions were followed during a median of 23 months to discard ma- benign lesions. 28% of the benign lesions showed both hypervascularization and lignancy. EC-MR diagnosed 26 out of 41 HCC nodules (sensitivity 63.4%, washout too. In lesions >10mm, the sensitivity/specificity of these features was CI95%: 46.9-77.9). The sensitivity and specificity for EOB-MR were 56.1% 75%/67%. Adding signal hyperintensity on T1-w images raised the specificity to (CI95%: 39.7-71.5) and 90.5% (CI95%: 69.8-98.8), respectively, with a positive 82%. predictive value (PPV) of 92 (CI95%: 74-99) and negative predictive value (NPV) Conclusion: Washout appearance was observed in close to one-third of benign of 51.4 (CI95%: 34.4-68.1). False positives were observed in the two ICC. In lesions leading to an unacceptably low specificity of this feature for the diagnosis EOB-MR, wash-in was present in 82.9% and portal wash-out in 42%. Low signal of HCC. Other imaging ancillary features and patients characteristics should be intensity in the 20-minute HBp was observed in 63.4% of HCC nodules, whereas carefully looked at, especially pre-contrast signal intensity and serum AFP level. the remaining 15 HCCs were iso- or hyperintense. Conclusion: EOB-MR is not superior to EC-MR for non-invasive diagnosis of SS 1.8 HCC in nodules <2 cm in cirrhotic patients. Liver imaging reporting and data system v2017 categorization of HCC using multiphasic MDCT: SS 1.6 comparison between patients with and without moderate Proliferation status of HCC predicted by whole-lesion to severe fatty liver 1 1 1 2 texture analysis on gadolinium ethoxybenzyl S.S. Kim , J.A. Hwang , H.C. Shin , S.-Y. Choi ; 1 2 diethylenetriamine pentaacetic acid-enhanced MRI Cheonan-si, Chungcheongnam-do/KR, Bucheon/KR Z. Ye; Chengdu/CN Purpose: To compare the sensitivity of the liver imaging reporting and data sys- Purpose: To explore the potential association between whole-lesion texture fea- tem (LI-RADS) v2017 for categorizing HCC using multiphasic MDCT between tures on gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB- patients with and without moderate to severe fatty liver (MSFL). DTPA)-enhanced MRI and the Ki67 labeling index (Ki67 LI) in HCC patients. Material and methods: This retrospective study was approved by the institu- Material and methods: 61 consecutive surgically confirmed HCC patients were tional review board, and the informed consent requirement was waived. A total prospectively examined on a 3.0 T scanner. Texture features on individual se- of 106 high-risk patients with 112 pathologically proven HCCs who underwent quence, including pre-contrast T1-weighted (T1-pre), T2-weighted (T2WI), arte- multiphasic MDCT were included. Patients were classified into MSFL (n=26) and rial phase (AP), portal venous phase (PVP) and hepatobiliary phase (HBP) imag- non-MSFL (n=80) group according to unenhanced CT liver and spleen parenchy- ing were derived using an in-house software (Analysis Kit, GE Healthcare). The mal attenuation. Two independent radiologists assigned LI-RADS categories and Ki67 LI was histopathologically determined and classified into low-proliferation accessed HCC features on MDCT. Sensitivities of LR-5/tumor in vein (TIV) and (Ki-67 LI<15%) and high-proliferation group (Ki-67 LI>15%). Lasso model was frequencies of major HCC features on MDCT were compared between two adopted for feature selection. Differences in texture features between low- and groups. high-proliferation groups were compared. Diagnostic performance of each se- Results: Sensitivities of LR-5/TIV were not significantly different between MSFL lected texture parameter in differentiating high-proliferation from low-prolifera- and non-MSFL group (73.1% [19/26] vs. 83.7% [72/86], P = 0.255; 80.8% [21/26] tion group was evaluated by receiver operation characteristic (ROC) analysis. vs. 87.2% [75/86], 0.522 in reviewers 1 and 2, respectively). No significant differ - Results: The cluster prominence and high grey level run emphasis of T1-pre, the ences in frequencies of arterial hyperenhancement, washout, and capsule were Haralick correlation and inverse difference moment (IDM) of T2WI, the IDM of AP, observed between the two groups (96.2% [25/26] vs. 98.8% [85/86], P = 0.412; the cluster shade and high grey level run emphasis of PVP and the correlation of 80.8% [21/26] vs. 89.5% [77/86], P = 0.308; and 53.8% [14/26] vs. 57% [49/86], HBP showed significant differences between low- and high-proliferation groups P = 0.778, respectively). Two HCCs (7.7% [2/26]) in MSFL group showed higher (P<0.05). The largest areas under ROC curve (AUCs) of T1-pre (high grey level attenuation compared with background liver on unenhanced image. run emphasis), T2WI (Haralick correlation), AP (IDM), PVP (high grey level run Conclusion: LI-RADS using MDCT showed comparable sensitivity for the diag- emphasis) and HBP (correlation) were 0.76 (CI[0.63, 0.86]), 0.70 (CI[0.57, 0.81]), nosis of HCCs regardless of MSFL. 0.69 (CI[0.56, 0.87]), 0.71 (CI[0.58, 0.82]) and 0.70 (CI[0.57, 0.81]), respectively, for identifying high-proliferation HCC. Correlation of HBP showed highest sensi- tivity (91.43%, CI[76.9%, 98.2%]) while IDM of AP showed highest specificity (96.15%, CI[80.4%, 99.9%]). Conclusion: Whole-lesion texture features on Gd-EOB-DTPA-enhanced MRI are promising in predicting and characterizing proliferation status of HCC. S659 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 The Liffey A SS 1.9 Pre-treatment estimation of liver function using T1 Scientific Session SS 2 mapping of gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid-enhanced MRI and normal liver volume Insights in imaging of rectal cancer in patients with HCC T. Duan, L. Cao, M. Wang, B. Song; Chengdu/CN SS 2.1 Purpose: This study aimed to prospectively evaluate liver function in patients The standard of MRI rectal cancer staging reporting in with HCC using T1 mapping before and after 20 minutes of gadolinium-eth- clinical practice: a case for standardization? oxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) injection and P.J. Brown, H. Rossington, D.J.M. Tolan; Leeds/UK normal liver volume, compared with indocyanine green (ICG) retention at 15 Purpose: Rectal cancer staging with magnetic resonance imaging (RCS-MRI) min (ICG R-15). allows accurate assessment of tumours. Significant variability exists in the Material and methods: 101 consecutive patients with HCC were included in content of reports. Template-style reporting can improve reporting standards, this study. T1 relaxation time of the livers before and 20 minutes after Gd-EOB- but its use is not widespread. Given the implications for treatment we have DTPA injection (pre-T1 and post-T1) was measured, and the reduction (Δ) and evaluated standard practice amongst radiologists to measure the current qual- reduction rate of T1 relaxation time (Δ%) were calculated. Volumes of normal ity of RCS-MRI reports. liver parenchyma were conducted and the products of T1 parameters and vol- Material and methods: 16 UK colorectal cancer multidisciplinary teams ume were calculated. ICG R-15 was also recorded. Liver function insufficiency (MDTs) serving a population over 5 million in Yorkshire were invited to submit a was defined as ICG R-15 >10. T1 parameters, volumes and the products were maximum of 10 consecutive RCS-MRI reports from January 2016 for each correlated with both ICG R-15 using Spearman rank correlation analysis. Re- radiologist participating in the colorectal MDT. All radiologists were subspe- ceiver operation characteristic (ROC) analysis of predicting the liver function cialists in GI imaging with training in RCS-MRI. Reports were compared to a insufficiency was performed for T1 parameters, volumes and the products. reference standard based on key data points in The Union for International Results: The post-T1 (r=0.450, P<0.001), Δ (r=-0.287, P=0.011), Δ% (r=-0.419, Cancer Control (UICC)-TNM 5 staging and other recognised factors influenc - P<0.001) volume (r=-0.374, P<0.001), product of Δ and volume (r=-0.442, ing case management. P<0.001) as well as product of Δ% and volume showed weak to moderate Results: 293 RCS-MRI reports were submitted from 32 of 42 (76%) eligible correlations with ICG R-15. In predicting the liver function insufficiency, the radiologists. 81 of 293 (28%) reports used a template. Template report usage area under the ROC of post-T1, Δ, Δ%, volume, product of Δ and volume and significantly increased recording of key data points versus non-template re - product of Δ% and volume were 0.767, 0.619, 0.722, 0.745, 0.791 and 0.856, ports for extra-mural vascular invasion (EMVI) status (98.8% v 53.3%, p < respectively. 0.001) and circumferential-resection margin (CRM) status (98.8% v 63.2%, p < Conclusion: A combination of T1 mapping and normal liver volume can help 0.001). Local tumour stage (97.5% v 92.0%, NS) and nodal status (98.8% v in evaluating liver function. 95.8%, NS) were usually reported with similar frequency. Conclusion: Non-template reporting in RCS-MRI excludes critical variables SS 1.10 that influence clinical decisions for rectal cancer. Standardised template re - Combining CT-liver perfusion and MRI with hepato- porting significantly improves collection of key data. specific contrast agent to increase diagnostic accuracy in patients with suspected HCC: work in progress SS 2.2 A.A. Hatzidakis, G. Kalarakis, K. Perisinakis, E. Chryssou, Multi-exponential T2* mapping distinguishes benign A. Papadakis, A. Karantanas; Heraklion/GR from malignant lymph nodes in rectal cancer patients: Purpose: Gadoxetic acid-enhanced MRI (GaE-MRI) is used for HCC surveil- an ex vivo and in vivo experiment 1 1 2 1 lance. We sought to compare the diagnostic accuracy of CT-liver perfusion I. Santiago , J. Santinha , A. Ianus , N. Papanikolaou , 1 1 1 2 (CTLP) plus GaE-MRI versus GaE-MRI alone for detection and characteriza- C. Matos , N. Shemesh ; Lisbon/PT, London/UK tion of suspected HCC lesions. Purpose: To investigate multi-exponential decay in multi-gradient-echo (MGE) Material and methods: Sixteen patients (all male, 12 cirrhotic) under HCC MRI for benign/malignant lymph node (LN) distinction in rectal cancer patients, surveillance before or after percutaneous HCC treatment, underwent GaE-MRI both ex vivo at 16.4T and in vivo at 1.5T. and CTLP (with a maximum interval of 6 weeks). In total, 16 pairs of CTLP and Material and methods: Ex vivo experiment: 33 benign and 32 malignant LNs GaE-MRI examinations were studied. GaE-MRI, was performed on a 1.5T- were retrieved from 11N+ rectal cancer patient specimens, preserved in 4% system (Siemens Vision-Hybrid). HCCs were characterized according to con- formaldehyde, moved to 1% phosphate buffered saline (PBS) for 24h, im- ventional MRI criteria. CTLP-maps of mean slope of increase were generated mersed in Flourinert and acquired in a 16.4T Bruker scanner. A fat-suppressed using a 128-CT-system (GE Revolution HD) and the lesions were depicted MGE acquisition was acquired: 50 TEs (echo time) starting at 1.6ms; 1.4ms based on a previously determined cut-off value. Diagnoses based on GaE-MRI interval; repetition time (TR)=1500ms; flip angle (FA)=50º; slice thickness alone and GaE-MRI plus CTLP were compared with digital subtraction angiog- 2 (ST)=0.3mm; in-plane resolution=0.1x0.1mm ; bandwidth=125000Hz; 25 aver- raphy (DSA). ages. In vivo experiment: 8 rectal cancer patients underwent MGE during stag- Results: Of the total 44 lesions identified (median diameter 20mm, range ing on a 1.5T Philips scanner: 32 TEs starting at 2.37ms; 2.37ms interval; 5-124mm), 19 were characterized as HCCs by DSA. GaE-MRI identified 16 2 TR=1519ms; FA=55º; ST=4mm; in-plane resolution=0.42x0.42mm ; true-positive, 19 true-negative lesions and misdiagnosed 6 false-positive and bandwidth=431.3Hz; 2 averages. Six patients underwent total mesorectal ex- 3 false-negative lesions, providing 84.2% sensitivity and 76% specificity. The cision without neoadjuvant therapy and of the LNs retrieved, 36 benign and 27 combination of GaE-MRI and CTLP identified 19 true-positive, 24 true-nega - malignant were matched to MGE images. Data from ex vivo whole-node vol- tive and 1 false-positive lesion increasing sensitivity and specificity to 100% ume of interest (VOI) and in vivo single-slice whole-node region of interest and 96%, respectively (p=0.0025). Another 12-mm nodule, characterized as (ROI) were used for analysis. Datasets were fitted using 1-, 2- and 3-compart - HCC by CTLP/GaE-MRI, was initially not detected by DSA, but was revealed ment T2* models. Models were compared based on Bayesian information cri- in a subsequent one, 6 months later. terion. Histogram analysis was performed for ex vivo and receiver operating Conclusion: Combination of GaE-MRI and CTLP may increase small (<15mm curves (ROC) were computed. Mann-Whitney U test was employed for param- diameter) hepatic nodule characterization accuracy, enabling more efficient eter comparison. patient selection for early and individualized loco-regional treatment. Results: 2-compartment model ranked first in both datasets. For ex vivo, sig - nificant differences were found in metrics derived from T2*a, T2*b, f and ΔΩ, the most discriminative being P75 of f(p=0.002). For in vivo, T2*a showed the highest specificity for malignancy (0.94) and T2*b showed the highest discrim - inative power (auROC=0.76). The specificity of T2*a, T2*b and their combina - tion exceeded that of reported conventional imaging criteria. Conclusion: Our results indicate multi-compartment T2* mapping may be of added value for LN staging in rectal cancer. S660 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 2.3 SS 2.5 Comparison of 8 diffusion MR models to distinguish The tram track sign: a new, highly specific and reliable benign/malignant mesorectal lymph nodes from rectal sign for the detection of complete response after cancer patients performed ex vivo at ultrahigh field neoadjuvant therapy in rectal cancer 1 2 1 2 A. Ianus , I. Santiago , D. Alexander , C. Matos , I. Santiago, M.J. Barata, N. Figueiredo, O. Parés, C. Matos; 2 1 2 N. Shemesh ; London/UK, Lisbon/PT Lisbon/PT Purpose: To assess how different compartment models capture the signal be- Purpose: To assess the diagnostic value of the tram track sign to predict path- haviour and differentiate benign from malignant lymph nodes (LNs) retrieved ologic/sustained clinical complete response (CR) after neoadjuvant therapy in from rectal cancer patients‘ specimens, using a rich diffusion MRI protocol, at rectal cancer. 16.4T. Material and methods: We retreived from our prospectively organized rectal Material and methods: Twenty-six benign and 32 malignant LNs originating cancer database all patients who underwent neoadjuvant therapy followed by from 11 consecutive N+ patients who underwent surgery without neoadjuvant restaging MRI (+)10 weeks post-radiotherapy, between October 2013 and therapy were preserved in 4% formaldehyde, moved to 1% phosphate-buff- March 2017. Two dedicated radiologists blindly and independently reviewed ered saline (PBS) 24h before acquisition, immersed in Flourinert and imaged in T2 and diffusion-weighted images and graded response according to tumor a 16.4T Bruker scanner. Stimulated echo acquisition mode diffusion-weighted regression grade (mrTRG) and diffusion weighted imaging (mrDWI) using an imaging was employed using 4 different b values, varying diffusion gradient in-house ordinal scale from 1-5. They also assessed the presence/absence of times and 6 different gradient directions for each parameter combination. Data the tram track sign (mrTTS) - a double, uninterrupted, markedly T2-hypoin- were normalized for each diffusion time and fitted to 8 different diffusion mod - tense line at previous tumour location. Endoscopic data at same timepoint, els. Averaged signal over whole-node and up to 10 individual LN regions of graded according to an ordinal scale from 0-4, was also retrieved. For statisti- interest (ROIs), as delineated by a dedicated radiologist, were used. Models cal analysis, primary endpoints were to compare the accuracy of endoscopy, were compared based on goodness of fit. The benign vs malignant differentia - mrTRG, mrDWI and mrTTS for the identification of sustained complete re - tion ability of each model was analysed using general linear model (GLM) with sponders (SCR), defined as patients with pathologic CR or clinical CR at 1 binomial distribution and receiver operating characteristic (ROC) curves were year. computed. Results: 47 patients considered eligible: 26 males, mean age 63.8y. 20 under- Results: ZeppelinSphere was the model that ranked first in most instances, went surgery. No significant differences were found for endoscopy. For ob - whereas the Ball model (equivalent to ADC) ranked last in most instances. For server 1, significant differences were found between mrTRG /mrTRG 1-2 3- whole-node analysis, Tensor and ZeppelinBall models best differentiated be- (p=0.03) and mrTTS mrTTS (p=0.02), and for observer 2, between mrDWI / 5 +/ - 1-2 nign from malignant LNs (area under ROC=0.78 and 0.77, respectively). For mrDWI (p=0.03) and mrTTS mrTTS (p<0.01). Inter-rater agreement was poor 3-5 +/ - individual ROIs, models with restriction (BallSphereSphere and Zeppelin- for mrTRG[0.15(p=0.14)], moderate for mrDWI[0.55 (p<0.01)] and very good for Sphere) yielded better results. mrTTS[0.83(p<0.01)]. mrTTS provided a sensitivity of 0.56/0.61 for observers Conclusion: Accounting for restricted diffusion improved the data fit of diffu - 1/2, respectively, and a specificity/PPV of 1 for both readers. sion MRI signal in LNs from rectal cancer specimens, which may provide more Conclusion: mrTTS has very high specificity, positive predictive value and in - specificity towards tissue microstructure. It also impacted lymph node differ - terrater agreement for the identification of SCR in rectal cancer. Its sensitivity entiation. is, however, low; therefore, it should be combined with other post-neoadjuvant therapy assessment tools. SS 2.4 T2-weighted signal intensity to predict complete and SS 2.6 good response after neoadjuvant chemoradiation Patterns of fibrosis on rectal MRI in clinical complete therapy in patients with rectal cancer responders undergoing wait-and-see after 1 2 1 R.A.P. Dijkhoff , S.G. Drago , J. Van Griethuysen , chemoradiotherapy for rectal cancer: correlation with 1 3 1 D.M.J. Lambregts , F.C.H. Bakers , R.G.H. Beets-Tan , functional outcome 1 1 2 3 1 1 2 M. Maas ; Amsterdam/NL, Monza/IT, Maastricht/NL T.N. Boellaard , D.M.J. Lambregts , A. Delli Pizzi , 1 3 1 M. Van Der Sande , B.J.P. Hupkens , J. Van Griethuysen , Purpose: To determine whether T2-weighted signal intensity (T2W-SI) before 1 1 1 1 G.L. Beets , R.G.H. Beets-Tan , M. Maas ; Amsterdam/NL, and after neoadjuvant chemoradiation therapy (CRT) can predict complete and 2 3 Chieti/IT, Maastricht/NL good response in patients with rectal cancer. Material and methods: 171 consecutive patients with rectal cancer treated Purpose: To study whether morphologic patterns of fibrosis on T2W-MRI in with neoadjuvant CRT underwent MRI before and 8-10 weeks post-CRT. The non-operated clinical complete responders after chemoradiotherapy (CRT) for primary tumour and remnant after CRT were manually delineated on T2W-MRI. rectal cancer are related to long-term functional outcome. Histogram analyses were performed on these volumes of interest with pyradi- Material and methods: N=68 patients were retrospectively analysed. All had omics. Extracted parameters were mean, median, standard deviation (SD), a sustained complete response and underwent MRI follow-up (FU) 3-6 month- range, minimum, maximum T2W-SI before and after CRT. Change between ly as part of wait-and-see program (median FU 34 months (range 16-62)). Mor- pre- and post-CRT T2W-SI parameters was calculated. Heterogeneity of T2W- phology of the rectal wall was assessed (2-reader consensus) on the T2W SI was assessed by the coefficient of variance (CoV=SD/mean). T2W-SI pa - FU-MRIs and classified according to 4 patterns: [1] no fibrosis (=normalised rameters were compared between complete (CR; ypT0), good responders rectal wall), [2] minimal fibrosis, [3] full-thickness fibrosis or [4] irregular/spicu - (GR; ypT0-1) and non-responders (NR; ypT2-4). Reference standard was ei- lated fibrosis. Long-term functional outcome was assessed with the Vaizey ther histology after surgery or a wait-and-see programme with at least 2 years score, a 0-24 point score on bowel function/incontinence, as the main out- of recurrence-free follow-up. come. Vaizey-scores were compared between the 4 patterns of fibrosis. Results: 36/135 patients had a CR and 49/171 a GR. Pre-CRT maximum SI Results: 5 patients showed no fibrosis, 45 minimal, 15 full-thickness and 3 ir - (941.62 vs. 1149.63, p=0.028), heterogeneity of the SI (0.265 vs. 0.291, regular/spiculated fibrosis. Mean Vaizey score was 1.6 for the patients with no p=0.017) and SI range (828.56 vs. 1074.52, p=0.009) were significantly lower fibrosis versus 4.5/3.9/4.0 for the patients with minimal/full-thickness/irregular in CR than in NR. Pre-CRT minimum SI was significantly higher in both CR fibrosis, respectively (P=0.557). For the patients (n=38) with a distal rectal tu - (113.06 vs. 75.10, p=0.017) and GR (117.02 vs. 69.46, p=0.001) compared to mour (≤3 cm from the anorectal junction), mean Vaizey score was 1.3 in the no NR. Mean difference in SI between primary and restaging was larger in both fibrosis group versus 4.7 for the minimal fibrosis and 6.5 for the full-thickness CR and GR compared to NR (CR: -14257 vs. NR: -105.44, p=0.114 and GR: fibrosis group (no patients had irregular fibrosis; P=0.198). 176.83 vs. NR: -87.72, p=0.001). Conclusion: Patients with a clinical complete response after CRT who show a Conclusion: The pre-CRT T2W-SI is significantly higher in CR and GR with a normalised rectal wall without fibrosis appear to have a better functional out - larger decrease in mean SI after CRT. These parameters could be a potential come in terms of incontinence and bowel function compared to patients with non-invasive marker for predicting complete/good response in patients with fibrosis. rectal cancer before and after CRT. S661 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 2.7 SS 2.8 18F-fluorodeoxyglucose positron emission tomography Ulcerated scars post neoadjuvant therapy in rectal ([18F]-FDG PET)/MRI in locally advanced rectal cancer cancer: can morphologic changes and diffusion- after preoperative chemo-radiotherapy: a comparison weighted imaging restriction patterns predict a with conventional imaging complete response? F. Crimì, C. Lacognata, I. Maretto, A. Perin, D. Cecchin, M.J. Barata, I. Santiago, N. Figueiredo, O. Parés, C. Matos; P. Zucchetta, F. Pomerri; Padua/IT Lisbon/PT Purpose: To assess the accuracy of restaging after preoperative chemoradio- Purpose: Are changes in angle, thickness and pattern of restriction to diffusion therapy (pCRT) with whole-body PET/MRI, pelvic T2-weighted (T2W) plus dif- of the tumour scar predictive of complete response in patients with rectal can- fusion-weighted imaging (DWI) MRI (pelvic-MRI), pelvic T2W MRI (standard- cer presenting with an ulcer at tumour site on endoscopy after neoadjuvant MRI) and thoracoabdominal CT when predicting histopathologic TNM stage in therapy (NAT)? locally advanced rectal cancer (LARC). Material and methods: All patients with rectal cancer who underwent NAT Material and methods: 26 patients with LARC underwent PET/MRI and CT and presented with an ulcerated scar on restaging endoscopy were retrieved before and after pCRT for TNM staging. 21 were treated with total mesorectal from our database. MRIs were reviewed independently and blindly by 2 radi- excision and 5 with transanal local excision. Histopathologic findings or a fol - ologists, which recorded, in 4 different timepoints post-radiotherapy: 1st low-up of at least 1-year were the reference standards. One radiologist evalu- (mean: 8,6 weeks), 2nd (mean: 20,7 weeks), 3rd (mean: 33,4 weeks) and last ated pelvic MRI and CT. A second radiologist evaluated standard MRI. A third (mean: 81,2 weeks): the angle between deepest point and inner borders of ul- radiologist and a nuclear medicine physician assessed PET/MRI. T staging cerated scar (ANG); the scar thickness (STh); and the distribution pattern of results were grouped in T0 and residual disease (T≥1). N stage was classified high-signal intensity on diffusion-weighted imgaing (DIST). Primary outcome on a per-patient basis as positive or negative using MRI dimensional criteria was presence/absence of signs of viable tumour at pathology or ≥1-year fol- (≥5mm per node), MRI lymph node global size reduction rate criteria (reduc- low-up (including regular digital rectal examination, endoscopy and MRI), pa- tion<70%) and PET/MRI dimensional criteria and/or nodal FDG uptake. tients considered non-complete responders (NCR) and complete responders Results: Sensitivity and specificity for ypT0 were 100%-85.7% for PET/MRI, (CR), respectively. Mann-Whitney-U and Fisher´s exact test were used for sta- 94.7%-85.7% for pelvic-MRI and 94.7%-57.1% for standard-MRI. For ypN+ tistical analysis. Interobserver agreement was assessed using intraclass cor- with dimensional criteria, sensitivity and specificity were 100%-88.9% for PET/ relation coefficient (ICC). MRI and 75%-88.89% for pelvic-MRI. The values for pelvic-MRI changed to Results: 15 patients considered eligible (mean age= 66y; 8 men), 7 underwent 87.5% and 72.2% using lymph node global size reduction rate criteria. PET/ surgery, 6 had persistent disease at pathology. Significant differences were MRI correctly diagnosed two liver and one distant nodal metastases while found for both readers in percentage change of ANG and STh (median values): missed a lung metastasis. ANG1st_ANG3rd: Reader1:NCR=+9,2%; CR=-24,7%, p=0.02; Reader2: Conclusion: PET/MRI improves the accuracy of ycTN staging compared to NCR=+44%; CR=-18,2%, p<0.01; ICC:0,44; ANG2nd_ANG3rd: Reader1: MRI, but performs worse than CT in ycM staging. Initial results are promising; NCR=+3,1%; CR=-16,2%, p=0.01; Reader2: NCR=+11,5%; CR=-13,8%, however, a larger cohort of patients should be examined introducing sequenc- p=0.04; ICC:0.34; ANG1st_ANGlast: Reader1: NCR=+15,8%; CR=-24%, es for lung and gadolinium for liver metastases. p=0.03; Reader2: NCR=+45,3%; CR=-30%, p=0.04; ICC:0.70; STh1st_ STh2nd: Reader1: NCR=0%;CR =-14,9%; p<0.01; Reader2: NCR=0%; CR=- 18,3%; p=0.04; ICC:0,80; STh2nd_STh3rd: Reader1: NCR=5,1%; CR=-9,8%; p<0.01; Reader2: NCR=+40%; CR=-6,3%; p=0.04; ICC:0,48; STh1st_STh3rd: Reader1: NCR=0%; CR=-29,6%; p=0.02; Reader2: NCR=22,2%; CR=-26,8%; p<0.01; ICC:0,96; DIST was significantly different for both readers at last time - point (absent/endoluminal linear vs intramural/irregular; p=0.01, ICC: 0.88). Conclusion: Patients with an ulcerated scar on endoscopy after NAT were more likely to be CRs when progressive contraction and thinning of the scar was observed on follow-up MRIs, and when an absent/endoluminal linear pat- tern of restriction to diffusion was observed at last follow-up. S662 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 Liffey Hall 2 SS 2.9 Endorectal ultrasound is accurate for the assessment of Scientific Session SS 3 anterior resection margin in low rectal cancer 1 2 1 1 1 J. Shur , D. Burling , A. Corr , J.T. Jenkins ; London/UK, Imaging of acute abdominal conditions Harrow/UK Purpose: Accurate assessment of anterior resection margin (ARM) in low rec- SS 3.1 tal cancer (LRC) helps ensure an R0 excision and appropriate triage to neo- Can volumes of extrapancreatic inflammation and adjuvant treatment (naCRT), and with MRI can be difficult due to paucity of pancreas on abdominal computed tomography predict mesorectal fat and presence of vessels posterior to Fascia of Denonvilliers. severity of acute pancreatitis? The purpose of this study is to compare endorectal ultrasound (ERUS) with E. Guler, T. Kose, M. Harman, O. Ozutemiz, N.Z. Elmas; MRI for prediction of involved ARM in LRC. Izmir/TR Material and methods: We identified rectal cancer patients who underwent Purpose: To evaluate the volumes of extrapancreatic inflammation and pan - ERUS over a 5-year period. We reviewed imaging reports and case notes of creas in predicting severity of acute pancreatitis compared with current scor- those with a LRC (within 5cm of anorectal junction) and threatened ARM re- ing systems and laboratory markers. ported on MRI in whom ERUS was undertaken for further evaluation. Histopa- Material and methods: This retrospective study included 30 patients (17 thology was used as a reference standard for the involvement of ARM and women, 13 men; median age 66.5 years) with acute pancreatitis who were sensitivity, specificity, positive predictive value (PPV) and negative predictive examined with computed tomography (CT) in early disease onset. Extrapan- value (NPV) were calculated for ERUS and MRI. creatic inflammation volume (EPIV), pancreatic volume (PV), ratio of EPIV to PV, Results: 24 patients had LRC with threatened ARM reported on MRI. Com- Balthazar score, CT severity index (CTSI), and modified CTSI were calculated. plete data including MRI, ERUS and subsequent histopathology report in 9 Amylase, lipase, C-reactive protein (CRP), and white blood cell (WBC) levels at patients were available, 5 who had naCRT. ERUS agreed with MRI for ARM the time of obtained CT scans were reviewed. Outcome parameters included involvement in 3 cases. 6 patients were appropriately triaged to more exten- the length of hospital stay and development of complications. Spearman’s sive resection (removal of posterior capsule of prostate or posterior vaginal rank correlation and Mann-Whitney U tests were used in the analysis of cor- wall) following confirmation of involved margin by ERUS, and 3 had less exten - relations. sive surgery following confirmation of a clear margin seen at ERUS. Sensitivity, Results: Significant positive correlations were found between EPIV and Balt - specificity, PPV and NPV was 80%, 75%, 80% and 75% respectively for ERUS hazar score, CTSI, and mCTSI (r:0.709, r:0.741, r:0.799, respectively and compared to 40%, 20%, 33% and 25% for MRI. p<0.001 for all). There were no significant correlations between pancreatic vol - Conclusion: ERUS is an accurate technique for complementary assessment ume and scoring systems. Positive correlations between EPIV/PV and 3 scor- of ARM in low rectal cancer. ing systems were depicted (p<0.001). Duration of hospitalization correlated with CTSI and mCTSI (r:0.47 and r:0.546, p<0.05). Significant correlations be - SS 2.10 tween the presence of complications and EPIV, EPIV/PV, CTSI, and mCTSI The diagnostic performance of diffusion-weighted MRI were observed (p<0.05). There were no significant correlations between amyl - and computed tomography in the detection of ase, lipase, CRP, WBC levels and 3 scoring systems. extramural venous invasion in rectal cancer Conclusion: EPIV and EPIV/PV can predict severity of acute pancreatitis. CTSI A. Gursoy Coruh, E. Peker, A. Erden; Ankara/TR and mCTSI calculated in early onset of disease correlate with length of hospital stay and complications. Purpose: To evaluate the diagnostic utility of diffusion-weighted magnetic resonance imaging (DWI-MR) and CT in the detection of extramural venous invasion (EMVI) in rectal cancer. SS 3.2 Material and methods: Out of 358 patients with rectal cancer, totally 58 pa- Changes in radiology workflow in the setting of tients (30 patients with mrEMVI score: 3-4 and ctEMVI score: 2-3 and 28 con- antibiotic treatment for uncomplicated appendicitis trol patients without EMVI) were enrolled in the study. Apparent diffusion coef- D.J. Bowden, M. Sheehan, A. Lee, R. Dunne, A. Hill, M. Lee, ficient (ADC) values of the tumor and EMVI(+) vein were measured. Diameter of M. Morrin; Dublin/IE superior rectal vein (SRV) and inferior mesenteric vein (IMV), distant metastatic Purpose: At a university teaching hospital, a single centre, prospective trial of spread were evaluated on CT. Pathology was accepted as the gold standard. the management of acute uncomplicated appendicitis (AUA), randomising pa- Results: Mean diameters of SRV (4.8±0.9mm vs. 3.6±0.8mm) and IMV tients into conservative treatment with antibiotics versus treatment with sur- (6.9±0.8mm vs. 5.4±0.9mm) were significantly larger (p<0.001) and ADC val - gery was undertaken (COMMA trial NCT#02916134). We analysed the pattern ues of the tumor and the vein were significantly lower (926.4±281.8 s/mm vs. of imaging requested in this group, and compared it with the pattern of imag- 1026.6±246.8 s/mm , p=0.032) in EMVI (+) patients compared to the control ing in patients who had undergone appendicectomy for clinically suspected group. A diameter of 3.95mm for the SRV (sensitivity: 93%, specificity: 66%, appendicitis during the year preceding the trial. The aim is to estimate the im- accuracy: 80%); 5.95 mm for the IMV (sensitivity: 93%, specificity: 71%, ac - pact on radiology workflow in a busy academic centre on transitioning from curacy: 82%) and 0.929 s/mm for ADC value was found as cut-off value by standard surgical treatment of AUA to antibiotic treatment. ROC analysis, for discrimination of EMVI (+) and EMVI (-) patients. When at Material and methods: Patients who underwent appendicectomy before the least two of these 3 criteria are present, sensitivity, specificity values and ac - COMMA trial and all patients enrolled in the COMMA trial were studied (pre- curacy increase (sensitivity: 100%, specificity: 75%, accuracy: 87%). The COMMA Nov 2014–Sept 2015, COMMA Sept 2015 – Dec 2017). The frequen- presence of distant metastases at presentation was significantly more preval - cy of positive imaging and histology was recorded in both groups. ant in EMVI (+) patients (p=0.002). Results: In pre-trial period, 282 appendicectomies were performed. This Conclusion: Measurement of ADC values and SRV-IMV diameters seems to group underwent 63 ultrasound scans (22%), 57 (20%) CTs and 1 MRI (0.3%). be a reliable method in the detection of EMVI in rectal cancer. EMVI (+) patients Negative histology at appendicectomy during the pre-trial period was 23% appear to have a higher risk for distant metastes at diagnosis. (n=64). During the trial period, 259 patients with clinically suspected appendi- citis were assessed. This group underwent 149 ultrasounds (56%), 78 CTs (30%) and 70 MRIs (27%). Of this group, 202 were randomized to antibiotic treatment and 198 to surgery. Negative histology at appendicectomy was 2.5% (n=5). Conclusion: Pre-operative imaging substantially reduces the number of nega- tive appendicectomies. There was a substantial increase in radiology tests ordered in the transition from surgical treatment of AUA to antibiotic treatment which will have considerable implications for radiology workflow and will re - quire significantly increased resources to be diverted to radiology in order to be implemented. S663 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 3.3 SS 3.5 Can CT accurately predict the presence of a single Is unenhanced CT-scan enough for the diagnosis and adhesive band in adhesive small bowel obstruction? the management of elderly patients with acute J. Sammon, J. Byrne, A. Okrainec, A. Hanbidge; abdominal pain in emergency department? Toronto, ON/CA M. Barat, A. Kiani, P. Calame, M. Lagadec, M. Zappa, V. Vilgrain, M. Ronot; Clichy/FR Purpose: Small bowel obstruction (SBO) is a common surgical emergency, with adhesions being the most frequent etiology (65-80%). Laparoscopic man- Purpose: To compare the accuracy and reproducibility of unenhanced and agement has become an established technique, with successful outcomes contrast-enhanced CT-scan for the diagnosis of patients over 75 years old higher in cases of single adhesive band (SB-SBO). Most patients have CT admitted in emergency department (ED) for acute abdominal pain (AAP). prior to surgery, however, the presence of a single adhesive band is usually not Material and methods: 208 consecutive patients over 75 years old consulting commented on in reports. the ED for AAP with a CT-scan with and without enhancement were retrospec- Material and methods: Patients undergoing laparoscopic surgery for adhe- tively included. Three readers read both exams in two different readings: 1) sive SBO 2006–2014 were eligible. 110 patients were identified. 14 patients unenhanced images only, 2) full exam. Sensitivity and specificity for diagnoses were excluded: 12 CT >3 days pre-surgery and 2 no CT features of SBO. Two were compared to the gold standard, defined as the diagnosis obtained after abdominal radiologists independently reviewed the CTs. A single band was a complete clinico-biological and radiological evaluation. Intra and inter-reader recorded for patients who had a single transition point or had two closely re- agreement were calculated for each reader using the kappa test and the inter- lated transition points. Other factors including the degree, presence of a class correlation. Sub-group analyses were performed for patients with an or- closed loop, free fluid, decreased perfusion, mesenteric oedema, bowel wall ganic disease as final diagnosis and for patients requiring an interventional thickening and pneumatosis. treatment. Results: 58 patients successfully underwent laparoscopic surgery, with 38 pa- Results: Diagnostic accuracy ranged from 64% to 68% without enhancement, tients requiring conversion to laparotomy. 33/58 laparoscopic patients had and from 68% to 71% after contrast enhancement. Contrast-enhanced imag- SB-SBO and 9/38 converted patients had SB-SBO. Inter-observer variability es did not significantly improve the diagnosis accuracy (P = 0.973 – 0.979). CT between the two readers for SB-SBO on CT was very good with κ=0.739 (95% scan corrected the diagnosis proposed by the ED physician in 59.1% (range CI: 0.579-0.899) and κ=0.807 (95% CI: 0.687-0.927) with the surgical result. 58.1-60.0%) and 61.2% (range 57.6-65.5%) of the patients before and after Reader 1 accurately assessed the presence of a SB-SBO in 40/42 cases (95% contrast injection (p>0.05). Intra-observer agreement between readings was sensitivity {95% CI: 0.888-1.017}) and Reader 2 in 37/42 cases (88% sensitiv- moderate to substantial (K=0.513-0.711). Inter-reader agreement was sub- ity {95% CI: 0.783-0.979}). stantial for both unenhanced (k=0.745-0.789) and full exam readings (k=0.745- Conclusion: CT can predict the presence of a single adhesive band and pre- 0.799). These results were similar in sub-group analysis. dict more complex cases, helping the surgical team triage patients appropri- Conclusion: Unenhanced CT-scan alone is accurate and reproducible enough ately to a laparoscopic approach, improving patient morbidity and length of for the diagnosis of patients >75 years old presenting with an AAP. hospital stay. SS 3.6 SS 3.4 Clinical utility of functional MRCP in the diagnosis of Increased unenhanced bowel-wall attenuation: a acute cholecystitis specific sign of bowel necrosis in closed-loop small- S. Ramanathan, D. Kumar, M. Heidous, A.A. Faki; Doha/QA bowel obstruction Purpose: To assess the feasibility and diagnostic accuracy of functional 1 2 1 1 C. Rondenet , I. Millet , L. Corno , I. Boulay-Coletta , MRCP (fMRCP) with hepatobiliary contrast in the diagnosis of acute cholecys- 2 1 1 2 P. Taourel , M. Zins ; Paris/FR, Montpellier/FR titis. Purpose: To identify computed tomography (CT) findings associated with Material and methods: 150 patients who underwent gadolinium ethoxybenzyl bowel necrosis in patients with surgically confirmed strangulating closed-loop diethylenetriamine pentaacetic acid (Gd-EOB-DTPA) enhanced MR imaging small-bowel obstruction (CL-SBO) due to adhesions or internal hernia. for various clinical indications were included. Study group comprised 60 pa- Material and methods: This retrospective study was approved by our institu- tients who presented with acute right upper quadrant pain and equivocal tional review board, and informed consent was waived. To identify CT signs of physical examination and/or ultrasound findings for acute cholecystitis. 90 bowel necrosis, two gastrointestinal radiologists performed blinded, inde- patients who came for other indications were considered as control group. pendent, retrospective reviews of 41 CT studies from consecutive patients Gallbladder was assessed for morphological features of acute cholecystitis. who had CL-SBO due to adhesions or internal hernias and who underwent Functional MRCP was evaluated for timing of contrast appearance in the bil- surgery within 48h. Based on surgical and pathological findings, patients were iary confluence, common bile duct (CBD), contrast reflux in to gall bladder classified as having reversible ischemia or histologically documented necrosis. (GB)/cystic duct and excretion into duodenum. Univariate statistical analyses were performed to assess associations between Results: In the control group, contrast reflux into GB/cystic duct occurred in CT signs and bowel necrosis. K statistics were computed to assess interob- 85/90 patients (94.4%) at 15 ± 4 minutes (range: 5-30 min). 5/90 patients who server agreement. did not show reflux demonstrated no extrahepatic biliary excretion due to poor Results: We included 25 (61%) women and 16 (39%) men with a median age hepatocellular uptake for other reasons. In the study group, 35/60 (58%) pa- of 79 years. Bowel necrosis was found in 25/41 (61%) patients and ischemic tients demonstrated reflux within 30 min. All these patients improved on clini - but viable bowel in 16/41 (39%) patients. Increased unenhanced bowel-wall cal and US follow up without acute cholecystitis. 25/60 patients did not dem- attenuation was the only CT finding significantly associated with bowel necro - onstrate contrast reflux till 1 hour delayed image. Acute cholecystitis was sis (P=0.0002). This sign had 58% (95%CI, 37-78) sensitivity and 100% (95% proven in 21/25 patients (84%) by cholecystectomy and development of con- CI, 79-100) specificity for necrosis. Interobserver agreement was fair (0.59; vincing morphological features of acute cholecystitis. 4/25 (16%) patients did 95%CI, 0.37-0.82). not demonstrate contrast reflux. Conclusion: Increased unenhanced bowel-wall attenuation is specific for Conclusion: fMRCP is highly sensitive (96%) and specific (94%) in the diag - bowel necrosis and should lead to prompt surgery for bowel resection. nosis of acute cholecystitis, provided there is optimal extrahepatic biliary ex- cretion. It is particularly helpful in cases of acute cholecystitis where the clini- cal and sonographic features are inconclusive. S664 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 3.7 SS 3.9 Role of imaging in the diagnosis of acute uncomplicated Emergency CT evaluation of pancreatic surgery appendicitis in an adult population in a large prospective complications in relationship to preoperative rectal/ clinical trial pharyngeal swabs status A. Lee, M. Sheehan, D. Bowden, P. O‘Leary, A. Hill, M. Lee, A. Fighera, G. Cardano, A.J. Cybulski, L. Bertuzzo, R. Dunne, M. Morrin; Dublin/IE G. Marchegiani, G.A. Zamboni, G. Mansueto; Verona/IT Purpose: Acute appendicitis is the most common cause of an acute abdomen Purpose: Multiresistant bacteria are becoming more and more common, and in adults. The purpose of this study is to assess the use of imaging in the diag- represent a cause of significant increase in morbidity and mortality in surgical nosis of acute uncomplicated appendicitis (AUA). patients. Our purpose was to analyze a series of patients who underwent Material and methods: Over an 18-month period, a prospective trial of pa- emergency CT for suspected abdominal complications after pancreatic sur- tients who met inclusion criteria for AUA were randomised into two treatment gery, correlating the findings with the preoperative rectal/pharyngeal swabs. limbs based on radiological findings on ultrasound+/-CT or MRI: conservative Material and methods: We searched our picture archiving and communica- management with antibiotic therapy or surgical appendicectomy. CT or MRI tion system (PACS) system for the first emergency postoperative CT scan for were performed if US was inconclusive. suspected abdominal complications after pancreatic surgery performed in our Results: 117 patients that were suspected of having AUA, met the inclusion institution in 1 year, thus obtaining 89 CT scans. Twenty-one patients had criteria and were randomised. 38.5% (N=45/117) had ultrasound alone prior to positive preoperative rectal or pharyngeal swabs, and 68 negative swabs. We randomisation. 19.6% (N=23/117) had MRI and 11.11% (N=13/117) had CT reviewed the scans to assess for the presence and type of complications. following a negative or inconclusive US. 38.5% were randomised after US Results: Among the 89 patients who underwent emergency CT, 21 (23.6%) alone, of which 53.33% (N=24/45) were randomised to the surgical arm. had positive swabs and 68 (66.4%) negative swabs. Patients with positive 79.16% (N= 19/24) had an appendicectomy with a pathologically proven AUA. swabs underwent CT on average 6.6 days after surgery, compared to 8.9 days Of note, negative histology rate was very low at appendicectomy 2.5% (N=5). for patients with negative swabs (p= 0.0021). The prevalence of patients with Discordant cases included 3 pathologically normal appendices, 1 of which had positive swabs was higher among the 28 patients with CT signs of bleeding: a faecolith at surgery; 1 gangrenous appendix (complicated appendicitis); and 28.6% patients with bleeding had positive swabs. For fistulas and collections, 1 granulomatous appendix (thickened but not inflamed). 46.6% (N=21/45) of the proportion of patients with positive or negative swabs was similar to that of whom had US alone were randomised to the conservative arm. Of these, the population who underwent these emergency CT scans. 90.47% (N=19/21) were successfully managed conservatively. 2 patients re- Conclusion: Patients who undergo pancreatic surgery with preoperative posi- turned for an appendicectomy. 1 patient had pathologically proven AUA and 1 tive swabs, undergo abdominal CT for suspected complications on average had a histologically normal appendix. 2.3 days earlier than patients with negative swabs. The prevalence of patients Conclusion: Traditionally, patients suspected of having AUA have been treat- with positive swabs is increased in patients with a CT positive for bleeding. ed with surgical resection. However, pre-operative imaging substantially re- duces the number of unnecessary negative appendicectomies, in addition to SS 3.10 identifying those patients who will respond to a conservative treatment with Computed tomography for acute abdominal antibiotics. haemorrhage: impact upon emergency embolisation A.C. O‘Brien, G.M. Healy, N. Rutledge, A. Patil, SS 3.8 J.W. McCann, C.P. Cantwell; Dublin/IE Acute jejunoileal diverticulitis: multicenter descriptive Purpose: To assess the utility of computed tomography (CT) prior to attempt- study of 33 patients ed embolisation of acute abdominal haemorrhage. 1 2 1 3 3 P. Lebert , I. Millet , O. Ernst , I. Boulay-Coletta , L. Corno , Material and methods: All patients who underwent attempted emergency 2 3 1 2 3 P. Taourel , M. Zins ; Lille/FR, Montpellier/FR, Paris/FR embolisation for acute abdominal haemorrhage at our institution between Purpose: Acute jejunoileal diverticulitis is a very rare and potentially serious 01/01/2010 and 31/12/2016 were included. Obstetric, variceal bleeds and disease involving mostly the elderly. The diagnosis is based on imagery but haemodynamically stable patients were excluded. 138 patients underwent 154 remains unrecognized. The purpose of this study is to describe the clinical and procedures, of which 131 were preceded by CT. Of the 154 procedures, suc- computed tomography (CT) features and the outcomes of acute jejunoileal cessful treatment was performed for active bleeding (74 cases), pseudoaneu- diverticulitis. rysm (26), AV shunt (6), irregular/truncated artery (10) and tumour (3). Empiric Material and methods: Cases of acute jejunoileal diverticulitis managed at treatment was performed in 20 cases. three French hospitals were identified retrospectively from 2005 through 2015. Results: Performing a CT before attempted emergency embolisation was as- The final diagnosis relied either on a clinical and radiological data review by a sociated with a significantly higher chance of successful treatment compared panel of experts, or on the surgical findings. Demographic, clinical, laboratory, to those with no CT (81% vs 59% respectively, p=0.023). For those who under- and 18-month outcome data were collected. CT scans were reviewed by hav- went CT, an acute vascular finding (active bleed, pseudoaneurysm, AV shunt, ing two radiologists reach a consensus about the inflammatory diverticulum, irregular/truncated artery) was associated with a higher chance of successful evidence of complications, and presence of other bowel diverticula. treatment compared to those with no acute vascular finding (85% vs 52%, Results: We identified 33 cases in 33 patients with a median age of 78 years, p=0.002). Patients with no acute vascular findings on CT (n=18), demonstrated including 30 (86%) in whom an inflammatory diverticulum was identified, at the active bleeding (5), pseudoaneurysm (2) and irregular/truncated artery (3) on jejunum (n=26, 87%) or ileum (n=4, 13%). Extra-intestinal gas was seen in 10 formal angiography. (30%) and fluid in 11 (33%) patients. Other small-bowel diverticula were visible Conclusion: Performing CT prior to emergency embolisation for acute ab- in all 33 patients. The diverticulitis was mild and resolved with non-operative dominal haemorrhage increases the probability of successful treatment. An treatment in 22 (67%) patients and severe in the remaining 11 (33%) patients, acute vascular finding on CT further increases the probability of success. How - of whom 8 required emergency surgery. ever, the absence of acute vascular findings should not exclude consideration Conclusion: Acute jejunoileal diverticulitis is a rare and usually non-serious for embolisation in the acute patient. condition that chiefly involves the jejunum. A detailed CT assessment may al - low non-operative treatment. S665 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 Wicklow Hall 2A SS 4.2 Diagnostic accuracy of standardised uptake value Scientific Session SS 4 (SUVmax) in predicting malignancy of supraclavicular Abdominal oncology lymph nodes from primary oesophageal cancer C. O‘Rourke, I. Welaratne, S. Cournane, L. McLoughlin, J.V. Reynolds, C. Johnston, N. Sheehy; Dublin/IE SS 4.1 Diagnostic value of dynamic perfusion MRI imaging in Purpose: The purpose of this study was to determine the diagnostic accuracy patients with local advanced rectal cancer in the and optimum cut-off value of SUVmax on positron emission tomography (PET) assessment of chemo-radiation treatment: relation to to predict malignancy of supraclavicular lymph nodes (SCLNs) in patients with tumor regression grade at histology oesophageal carcinoma. S.G. Drago, A. Pecorelli, C. Talei Franzesi, M. Cressoni, Material and methods: All diagnosed cases of oesophageal cancer were ret- S. Sironi, D. Ippolito; Monza/IT rospectively reviewed (2012-2016). Patients that had a confirmed diagnosis of oesophageal cancer with avid SCLNs on staging PET were included in the Purpose: To investigate the value of dynamic contrast-enhanced perfusion- study. 33 SCLNs that subsequently underwent ultrasound guided biopsy for MRI parameters in the evaluation of the response to chemo-radiation therapy staging were analysed. The maximum uptake values (SUVmax) of the SCLNs in patients with local advanced rectal cancer in comparison with histology. and primary tumours were measured. A receiver operating characteristic Material and methods: Fourteen patients with biopsy proven rectal adeno- (ROC) analysis was performed to determine the optimum cut off of SUVmax in carcinoma who underwent dynamic contrast-enhanced MR performed on predicting malignancy. 1.5T MRI system (Achieva, Philips), before (MR1) and after chemo-radiation Results: 24/33 PET-detected SCLNs were malignant. ROC analysis identified therapy (MR2), were enrolled in this study. The protocol included T1 gadolini- the best nodal SUVmax cut-off to be 3.0. The diagnostic accuracy of PET was um-enhanced THRIVE sequences acquired on axial planes. A dedicated work- 76.0% (sensitivity = 78.9%, specificity = 66.6%). For SCLNs with SUVmax > station was used to generate color permeability maps; region of Interest (ROI) 3.0, PET showed a positive predictor value (PPV) of 88.2%; for SCLNs < 3.0, was manually drawn on tumor tissue and normal rectal wall, hence the follow- PET showed a negative predictor value of 50%. The median SUVmax of path- ing parameters were calculated and statistically analyzed: Relative venous ologically negative and positive nodes were 2.8 (range 1.8 - 6.0) and 5.3 (range enhancement (RVE), maximum enhancement (ME) and time to peak (TTP). 1.9 -13.4). The median primary tumour SUVmax was 13.8 (range 3.7 - 30.0). Perfusion parameters were related to pathologic tumor regression grade (Man- The SUVmax of metastatic lymph nodes were significantly higher than those of dard’s criteria; TRG1=complete regression, TRG5=no regression). benign lesions (p < 0.05). Results: Five tumors (36%) showed complete or subtotal regression (TRG1-2) Conclusion: Our study revealed an accuracy rate of 76% for PET detected at histology and classified as responders; 9 tumors (64%) were classified as SCLNs in patients with oesophageal carcinoma. For SCLNs with SUVmax > non-responders (TRG3-5). Perfusion MRI parameters were significantly higher 3.0, PET had a high PPV (88%), which can minimize the need for further diag- in the tumor tissue than healthy tissue in MR1 and MR2 (p<0,05). At baseline nostic tests. (MR1), no significant difference in perfusion parameters was found between responders and non-responders. After chemo-radiation therapy, at MR2, re- sponders showed significantly (p< 0.05) lower perfusion values (RVE(%)76±25; SS 4.3 ME(%):96±41;TTP(sec):13±26) compared to non-responders (RVE(%):153±50; Dedicated DW-MR imaging as a selection tool for ME(%):169±42;TTP(sec):38±68). Moreover, in responders, perfusion values hyperthermic intraperitoneal chemotherapy in patients decreased significantly at MR2 (RVE(%)76±25; ME(%):96±41;TTP(sec):13±26) with peritoneal carcinomatosis from colorectal origin compared to MR1 (RVE(%):118±15;ME(%):120±18;TTP(sec): 14±24) (p<0,05). I. Van‘t Sant, M. Engbersen, H. Van Eden, Furthermore, in non-responders, there was no significant difference between D.M.J. Lambregts, N.F.M. Kok, A.G.J. Aalbers, perfusion values at MR1 and MR2. R.G.H. Beets-Tan, M.J. Lahaye; Amsterdam/NL Conclusion: Dynamic contrast perfusion-MRI analysis represents a comple- Purpose: Combined cytoreductive surgery (CRS)/hyperthermic intraperitoneal mentary diagnostic tool for identifying vascularity characteristics of tumor tis- chemotherapy (HIPEC) treatment gives a subset of patients with peritoneal sue in local advanced rectal cancer, useful in the assessment of treatment re- carcinomatosis (PC) from colorectal origin (CRPC) with a limited and resecta- sponse. ble peritoneal tumor load and prospect of long-term survival or even cure. A non-invasive imaging method to accurately select potential CRS/HIPEC can- didates preoperatively is highly demanded. We compared the peritoneal can- cer index (PCI) estimated preoperatively by diffusion-weighted MR imaging (DW-MRI) with PCI found at surgery to assess whether DW-MRI can be used to select CRS/HIPEC candidates. Material and methods: Forty-nine consecutive patients (M/F=23/29) with confirmed or suspected PC from histologically proven colorectal origin were included. All patients underwent preoperative dedicated whole-body DW-MRI before exploratory laparoscopy or CRS/HIPEC. PCI was prospectively as- sessed by two independent radiologists on MR-imaging and was compared to PCI found at surgery (reference standard). Based on PCI patients were catego- rized as operable (PCI 0-20) or inoperable (PCI 21-39). Results: Mean PCI at surgery was 12.2 (range 0-34). Mean radiological PCI for reader 1 was 10.9 (range 2-33) and 9.2 (range 0-32) for reader 2. Radiologist 1 and 2 staged respectively 46/49 and 44/49 patients correctly (accuracy 94%/90%). Both radiologists detected all operable patients with a PCI<21 at surgery (sens=100%). Respectively 3 and 5 patients were understaged by DW-MRI. No patients were overstaged. The intraclass correlation between the radiologists was excellent (ICC=0.91, 95% CI: 0.758-0.959). Conclusion: In a large cohort, DW-MRI seems to be an accurate and robust selection tool to noninvasively select CRPC patients in which CRS/HIPEC is feasible. Further multicenter studies must be performed to see whether DW- MRI may replace surgical PCI staging. S666 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 4.4 SS 4.6 Radiological considerations in multiple site CT-liver-perfusion dose reduction protocols for cytoreductive surgery for metastatic colorectal cancer: detection of hepatocellular carcinomas: is it achievable? a cohort study A.A. Hatzidakis, G. Kalarakis, K. Perisinakis, N. Kosidekakis, A.D. Stirling, A.N. Murphy, C.S. McQuade, J. Mulsow, N. Kokkinos, A. Karantanas; Heraklion/GR C. Shields, H. Fenlon, C.G. Cronin; Dublin/IE Purpose: To compare the diagnostic efficiency of standard “35-passes” CT- Purpose: Cytoreductive surgery (CRS) can improve survival in selected pa- liver-perfusion (CTLP) acquisition protocol versus two “18-passes” CTLP-pro- tients with metastatic colorectal cancer. These are complex, major surgeries tocols for hepatocellular carcinoma (HCC) detection. that should be undertaken in specialist centres. The aim of this study is to Material and methods: Fourteen patients (13 male, age 57-87) with 32 con- evaluate outcomes relevant to radiological practice and resource management firmed HCCs (median diameter 25mm, 11-125mm) underwent CTLP using a in patients undergoing multiple site CRS for metastatic colorectal cancer. high temporal resolution acquisition protocol with 35 passes, pass-duration Material and methods: Patients who underwent multiple site CRS were iden- 1.7 sec, and total duration 59 sec. Eleven different perfusion maps were gener- tified from a prospectively maintained database of all patients with metastatic ated. Regions of interest (ROIs) were positioned on non-tumorous parenchyma colorectal cancer referred to a national centre for the management of colorec- and on HCCs. Maps were also retrospectively generated using the 18 first tal cancer and peritoneal malignancy. The number and anatomical location of passes (total acquisition time 31 sec) and the 18 odd numbered passes (total the operative sites, peritoneal carcinomatosis index (PCI) and completeness of acquisition time 59 sec). Receiver operating characteristic (ROC) analysis was cytoreduction were recorded from electronic medical records. The number of employed to evaluate the ability of each map to discriminate HCCs from non- multidisciplinary discussions, the extent of follow-up imaging and the number tumorous liver parenchyma. Comparison of ROC curves was performed to of image-guided interventions were recorded from a national picture archiving evaluate statistical significance of differences in the discriminating efficiency of and communication system (NIMIS PACS). the derived perfusion maps between different acquisition protocols. Results: 72 patients were included over a 3-year period from May 2014-2017. Results: Hepatic arterial blood flow, mean slope of increase, time to peak, The mean number of operative sites was 3 (range 2-7). The most common positive enhancement integral, and hepatic arterial fraction were found to be sites of CRS were: peritoneum and omentum (43%), liver (38%), and female the prevailing parameters regarding HCC discrimination when the “35-passes” reproductive organs (22%). This cohort had extensive imaging requirements protocol was used, with estimated area under ROC curve of 1.000, 0.995, averaging 10.5 investigations (either CT, MRI or positron emission tomogra- 0.993, 0.990, 0.983 respectively, without significant differences in their dis - phy/CT, excluding initial staging) and 4.4 multidisciplinary discussions. The criminating power (p>0.05), followed by Tmax, mean transit time (MTT), per- mean number of image-guided interventions was 2.4 (range 0-16). meability–surface area product (PS), blood flow (BF), blood volume (BV), Conclusion: The radiology input to the treatment paradigm of those undergo- IRFT0. Comparison of ROC curves between “35-passes” and “18-passes” ing CRS for metastatic colon cancer is substantial. Awareness of this will help protocols yielded no significant differences (p>0.05) for any of the studied guide resource management in this complex patient cohort. maps. Conclusion: Dose reduction in CTLP for HCC detection can be safely achieved by reducing the number of passes to 18, since the diagnostic accuracy of the SS 4.5 produced perfusion maps is not downgraded. Role of gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid MRI in characterizing nodules with atypical enhancement on CT: results of a prospective SS 4.7 monocentric study Peri-hilar cholangiocarcinoma preoperative CT and MRI: R. Faletti, M. Gatti, E. Caramia, A.M. Bozzato, A. Ferraris, accuracy and inter-observer agreement 1 1 1 1 S. Fiore, L. Bergamasco, P. Fonio; Turin/IT A. Sarno , A. Contro , A. Ruzzenente , S. Conci , 1 1 2 2 G. Tedesco , V. Ciaravino , R. De Robertis , N. Cardobi , Purpose: To assess the use of gadolinium ethoxybenzyl diethylenetriamine 1 1 2 M. D‘Onofrio ; Verona/IT, Peschiera del Garda/IT pentaacetic acid (Gd-EOB-DTPA) MRI in patients with “atypical” HCC diag- nosed at multiphase CT. Purpose: To evaluate the accuracy of CT and MRI in the preoperative staging Material and methods: 51 cirrhotic patients with at least one “naïve” lesion of peri-hilar cholangiocarcinoma. showing atypical dynamics enhancement on multiphase CT underwent Gd- Material and methods: CT and MRI study of 60 patients (45 males and 15 EOB-DTPA MRI were prospectively enrolled. Images were reviewed by two females, mean age 67 years) with peri-hilar cholangiocarcinoma were retro- radiologists in consensus. The atypical lesions were re-categorized by MRI spectively reviewed by two radiologists. All patients underwent surgical evalu- into: group A) typical HCC; group B) hypervascular without washout and hy- ation. The review of imaging data was performed blinded to surgical and pointense in the hepatobiliary phase (HBP) and group C) hypovascular nod- pathological results and separately. Involvement of biliary systems and vessels ules, hypointense in the HBP, hyperintense on diffusion-weighted imgaing was judged. Sensitivity, specificity and accuracy were calculated with respect (DWI). Patients were classified according to the Barcelona Clinic Liver Cancer to surgical and pathological results. Inter-observer agreement was calculated (BCLC) guidelines prior and after the MRI. Data were analysed using non-par- with K agreement test. ametric tests. Results: 50 patients were resected with curative intent (R0/R1) and 10 patients Results: 15 typical and 75 atypical lesions were found at CT. Twenty-six (66%) underwent explorative/palliative surgical treatment. The right bile ducts were atypical lesions were re-classified into in group A, 8 (21%) in group B and 5 involved in 31 patients, the left in 42 and both were involved in 26. Portal vein (13%) in group C; among the atypical lesion 17 were not confirmed, 12 were and hepatic artery involvement were present respectively in 36,2% and 23,2% not detected and the diagnosis remained uncertain in 7 at MRI. Typical lesions of patients. Lymph-node involvement was observed in 25 patients. The sensi- resulted bigger than atypical lesions (22.3±16.9 vs. 12.3±7.3 mm; p=0.0004) bility, specificity and accuracy of MR and CT were similar. Accuracy for CT was and in particular, a lesion >15mm had a positive likelihood ratio (LR+) of 2.25 76.4%, 79.1% and 55.1% for biliary, vascular and lymph-node involvement. to be typical. BCLC stage was changed after MRI in 61% (31/51) patients. For MR accuracy was 74.0%, 82.2% and 51.6%, respectively. Inter-observer Based on MRI a biopsy was indicated for 7 nodules compared to 43 based on agreement (K value) was higher for MR compared to CT in biliary involvement CT and 75% of group C lesions were confirmed histologically as HCC. (0.57 and 0.41), lower for arterial involvement (0.41 and 0.44) and lymph-node Conclusion: Gd-EOB-DTPA could correctly classify the majority of small and involvement (0.18 and 0.22). The agreement of the two readers showed statis- “atypical” HCC found on CT with high clinical impact on BCLC staging and tically significant differences (p<0.001) for the right lobe. management decision. Conclusion: CT and MRI are accurate in peri-hilar cholangiocarcinoma stag- ing. MRI is superior to judge biliary systems involvement. Inter-observer agree- ment was inferior for the right lobe. S667 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 4.8 SS 4.10 Improved delineation of hepatic metastases from Comparison of gadobutrol and gadofosveset trisodium colorectal cancer using noise-optimized virtual in MRI for rectal cancer 1 2 3 monoenergetic reconstructions from dual-energy R.A.P. Dijkhoff , J. Santinha , S.G. Drago , 1 1 4 computed tomography D.M.J. Lambregts , J. Van Griethuysen , F.C.H. Bakers , 1 1 1 2 1 2 1 J.L. Wichmann , M.H. Albrecht , S.S. Martin , L. Lenga , R.G.H. Beets-Tan , N. Papanikolaou , M. Maas ; 1 1 1 1 1 2 3 4 D. Leithner , B. Bodelle , T.J. Vogl , R. Hammerstingl ; Amsterdam/NL, Lisbon/PT, Monza/IT, Maastricht/NL 1 2 Frankfurt am Main/DE, Essen/DE Purpose: To compare the pharmacokinetics of gadobutrol (GDB; micromo- Purpose: To assess the impact of noise-optimized virtual monoenergetic im- lecular contrast-agent) with the albumin-binding contrast-agent gadofosveset aging (VMI+) reconstructions on quantitative and qualitative image parameters trisodium (GDF) in MRI for staging or rectal cancer using semiquantitative in patients with hepatic metastases from colorectal cancer at abdominal dual- analyses. energy computed tomography (DECT). Material and methods: 12 patients with primary rectal cancer were included. Material and methods: Forty-two patients (29 men; 67.1±12.3 years) with 6 patients underwent dynamic contrast-enhanced-MRI (DCE-MRI) with GDF histopathologically-confirmed hepatic metastases from colorectal cancer un - and were matched with 6 patients who had DCE-MRI with GDB. DCE-MRI derwent triphasic contrast-enhanced abdominal DECT. Images from the por- protocol was the same in all patients. Regions of interests were manually tal-venous phase were post-processed with the standard linear blending tech- placed in the tumour and DCE-MRI was processed with OncoAnalysis Horos nique (M_0.6) as well as with the traditional virtual monoenergetic (VMI) and Plugin (Computational Clinical Imaging Group, Champalimaud). Semiquantita- novel VMI+ algorithms in 10-keV intervals from 40 to 100 keV. Attenuation tive parameters (wash-in-slope, wash-out-slope, area-under-the-curve (AUC), measurements were performed in hepatic lesions, unaffected liver tissue, rec- maximum enhancement (MaxEnh) and time-to-peak (TTP)) based on signal tus abdominis muscle, and adjacent visceral fat to calculate objective signal- intensity time curves were extracted and compared between GDB and GDF. to-noise (SNR) and contrast-to-noise (CNR) ratios. Five-point scales were Using the coefficient of variance (standard deviation/mean) heterogeneity was used to evaluate overall image quality and lesion delineation by three blinded calculated for all semiquantitative parameters. radiologists with varying levels of experience. Results: Contrast-uptake parameters such as mean wash-in-slope (19.86 vs. Results: SNR and CNR (negative values indicating best contrast of hypoat- 22.16, respectively, p=0.690) and mean maximum enhancement (283.72 vs. tenuating lesions) of liver metastases were best in 50-keV VMI+ series (3.5±2.1 286.86, respectively p=0.615) were comparable between patients with GDB and -3.6±2.0), significantly superior to all other reconstructions (all P<0.001; and GDF, with only slight differences between both contrast agents. Also, SNR M_0.6: 2.4±1.5; CNR M_0.6: -2.6±1.8). Qualitative image parameters maximum wash-in-slope (58.71 vs. 57.88, p=0.974) was comparable for both showed highest values for 50-keV VMI+ reconstructions (median 5, respec- contrast agents. The differences between GDB and GDF for maximum/mini- tively; P≤0.023) regarding overall image quality. Qualitative assessment of le- mum/mean washout-slope were remarkably small, given the difference in size sion delineation peaked in 40-keV VMI+ (median 5) and 50-keV VMI+ (median of the agents. GDF led to a more homogeneous maximum enhancement than 4; P=0.067), significantly superior to all other reconstructions (all P<0.001; GDB (0.55 vs. 0.33, respectively, p=0.079). M_0.6: median 3). Conclusion: The albumin-binding contrast agent GDF shows similar contrast- Conclusion: Novel VMI+ reconstructions at 50 keV can substantially increase uptake parameters compared to micromolecular GDB in this small sample of quantitative image quality and improve subjective assessment of image quality rectal tumours. It seems that GDF behaves similarly to a micromolecular agent and lesion delineation of hepatic metastases from colorectal cancer compared shortly after injection, which might be explained by the absence of binding to to standard image reconstruction and traditional VMI. albumin immediately after injection. Possibly, results of DCE-MRI with GDF might be extrapolated to micromolecular agents, regarding wash-in phase. SS 4.9 How are we doing? Radiologist and multi-disciplinary team clinician opinions on the quality of MRI rectal cancer staging reports P.J. Brown, H. Rossington, D.J.M. Tolan; Leeds/UK Purpose: Rectal cancer staging with magnetic resonance imaging (RCS-MRI) allows accurate assessment of tumours. Significant variability exists in the content of reports. Colorectal multi-disciplinary teams (CRC-MDTs) make treatment decisions based on these reports. We evaluated the satisfaction of CRC-MDT members with current standards of radiology reporting to identify areas for improvement. Material and methods: 16 UK CRC-MDTs serving a population over 5 million in Yorkshire, were invited to complete an online questionnaire assessing opin- ions on RCS-MRI report content and clarity. This included questionnaires for RCS-MRI radiologist ‘reporters’ and other CRC-MDT ‘users’ who use the re- ports to plan care (including colorectal surgeons, oncologists, pathologists and nurse specialists). 21 of 42 (50%) ‘reporters’ and 48 ‘users’ completed the survey. All RCS-MRI reporting radiologists were subspecialists in GI imaging. Results: Questionnaires indicated items which were ‘always’ contained in re- ports, with tumour stage, nodal status and CRM status included in more than 80% of cases. However there was lower than expected reporting for tumour distance from the anal verge (81% ‘reporters’ v 65% ‘users’), relationship to peritoneal reflection (29% ‘reporters’ v 46% ‘users’), extramural vascular inva - sion (57% ‘reporters’ v 67% ‘users’) and inter-reporter consistency (76% ‘re- porters’ v 75% ‘users’). There was no statistically significant difference be - tween the opinions of reporters and users. Despite this 92% of ‘users’ were satisfied or very satisfied (on a five-point scale) with RCS-MRI. Conclusion: CRC-MDT users were satisfied with RCS-MRI reports despite inconsistencies for reporting critical variables. CRC-MDT user satisfaction of reports may not act as a sufficient stimulus to improve standards of RCS-MRI reporting. S668 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 11:00 - 12:30 Wicklow Hall 2B SS 5.2 MRI for response assessment after neoadjuvant Scientific Session SS 5 chemoradiotherapy in oesophageal cancer: added value Updates in upper and lower GI tract disor- of diffusion-weighted imaging S.E. Vollenbrock, F.E.M. Voncken, D.M.J. Lambregts, ders M. Maas, J.M. Van Dieren, L.C. Ter Beek, B.M.P. Aleman, R.G.H. Beets-Tan, A. Bartels-Rutten; Amsterdam/NL SS 5.1 Purpose: High accuracy is needed to select oesophageal cancer patients with Post chemoradiotherapy assessment of anal carcinoma a complete response (CR) after neoadjuvant chemoradiation (nCRT) for organ- using MRI: comparison with positron emission preserving treatment instead of oesophagectomy. Performance of T2-weight- tomography in biopsy proven patients ed MRI (T2W-MRI) and additional diffusion-weighted MRI (T2W+DW-MRI) was R. Faletti, M. Gatti, V. Carrozzo, L. Bergamasco, P. Fonio; determined. Turin/IT Material and methods: Thirty-seven patients with locally advanced oesopha- Purpose: To assess the use of MRI in local response assessment and detec- geal cancer underwent MRI (1.5 Tesla; T2W-MRI and DW-MRI, b-val- tion of early local relapse after chemoradiotherapy (CRT) in patients with anal ues=0,200,800 s/mm ) before and after nCRT with a maximum of 21 days be- squamous cell carcinoma (AC). tween MRI and surgery. Three radiologists initially scored T2W-MRI only. Material and methods: Eight patients with histologically proven AC who com- Thereafter they rescored T2W with DW-MRI. A 5-point score was used (1=def- pleted pre-(t0), post-CRT 4-week(t1) and 20-week(t2) MRIs and pre-(t0) and inite CR, 3=inconclusive, 5=definite residual tumour). Histopathology after oe - post-CRT 20-week(t2) positron emission tomography (PET) scan were pro- sophagectomy represented the reference standard (Mandard tumour regres- spectively included. Images were reviewed by two radiologists in consensus, sion grade 1=CR, 2-5=residual tumour). Area under the receiver operating blinded to clinical outcome: tumor size (D-mm), apparent diffusion coefficient characteristic (AUROC), sensitivity and specificity were calculated considering (ADC-mm /s), time to peak (TTP-sec) and standard uptake value (SUV) were MRI scores 3-5 as tumour-positive. recorded. Lesion biopsy performed at 6-month was the gold standard test for Results: Three out of 37 patients were excluded due to poor image quality. evaluation of response to treatments. Data were analyzed using chi-squared Seven (21%) of the remaining 34 patients achieved a CR. AUROCs were 0.57, test, Wilcoxon signed-rank test and Pearson’s correlation coefficient. 0.66 and 0.66 for T2W-MRI only and 0.75, 0.70 and 0.71 for T2W+DW-MRI Results: All patients were classified as responders. ADC, TTP and D were (p=0.09, 0.32, 0.70). Sensitivity for detecting residual tumour was 100%, 89% significantly different from t0 to t1 (0.8±0.2vs.1.2±0.2; p=0.002; and 89% for T2W-MRI only and 93%, 96% and 85% for T2W+DW-MRI. Spec- 95±55vs.191±59;p=0.03 and 31±14.5vs.18±17;p=0.002) and from t0 to t2 ificity improved from 14%, 43% and 43% on T2W-MRI to 57%, 43% and 57% (0.8±0.2vs.1.4±0.2;p=0.001, 95±55vs.214±67;p=0.006 and for T2W+DW-MRI. 31±14.5vs.10.5±10;p=0.001). From t1 to t2 ADC and TTP were not different Conclusion: T2W+DW-MRI is promising for response assessment after nCRT (p=0.12 and p=0.50) unlike D (p=0.04). A difference from t0 to t1 of 20% in in oesophageal cancer patients. Sensitivity is high, indicating that the risk of ADC (increase), TTP (reduction) and D (reduction) was identified as forecast of missing residual tumour is minimal. Specificity is lacking, indicating that the responder. SUV was significantly different from t0 to t2 (9.9±4vs.1.2±2.1;p=0.02). degree of residual tumour is overestimated. A multimodality approach with ADC was strongly inversely correlated with SUV values (r= -0.74). endoscopy, endosonography and positron emission tomography (PET)-CT Conclusion: D, ADC and TTP summarized in “the-rule-of-20%” seemed to be may further improve response assessment. a good marker of early (4-week) response to treatment. The correlation ADC/ SUV suggested an association between tumor cellularity and metabolic activ- SS 5.3 ity. ADC had a stronger relationship with outcome and may be more useful for Prognostic value of positron emission tomography predicting the prognosis of AC than SUV. (PET)-CT in oesophageal cancer patients undergoing oesophagectomy E. O‘Malley, C. Waters, C. Collins, J. Bruzzi; Galway/IE Purpose: Above the age of 50 oesophageal cancer incidence is 30-40/100,000 in the Irish Health Service Executive Western area. Stage I disease has 60% overall survival at 5 years, stage II 30%, stage III 20% and stage IV less than 5%. Accurate staging is important when selecting surgical candidates and multiple modalities are used including CT, endoscopic ultrasound and PET CT. PET CT may provide extra prognostic value which is not used in clinical prac- tice. Conflicting data exist about the usefulness of maximum standardised uptake value (SUV ) and other measurements such as metabolic tumour vol- max ume (MTV) and total lesion glycolysis (TLG). This study analyses PET CT find - ings and compares tumour measurements including MTV and TLG with the post-surgical mandard response score, 5-year survival and histological stag- ing. Material and methods: This retrospective analysis looked at PET CT studies from 2010 to 2016. Patient demographics, survival time, histopathology stag- ing and PET CT measurements were collected. PET CT measurements includ- ed SUV , SUV , SUV MTV and TLG. max peak mean Results: We identified 64 patients who underwent oesophagectomy. Median age at surgery was 64 (39-78). Histopathology was adenocarcinoma 72% and squamous cell carcinoma 28%. Mean MTV was 47.9 ml, mean SUV 9.9, max mean SUV 7.95, mean SUV 6.23 and mean TLG 207.79. Mean 5-year peak mean survival was 1.5 years. There was no significant correlation between PET CT measurements and survival, mandard score or histological staging. Conclusion: FDG avid measurements alone are not a useful prognostic factor in oesophageal cancer patients undergoing surgery. S669 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 5.4 SS 5.6 Restaging of resectable esophageal cancer using An investigation of the association between sarcopenia positron emission tomography (PET)/CT after and post-operative morbidity and mortality in patients neoadjuvant treatment: combining the best of both with gastric cancer using CT worlds S. O‘Brien, M. Twomey, F. Moloney, R.G. Kavanagh, D. Tamandl, B. Füger, A. Haug, A. Ba-Ssalamah; Vienna/AT M. Maher, O.J. O‘Connor, C. O‘Suilleabhain; Cork/IE Purpose: To assess the feasibility and diagnostic performance of PET/CT in Purpose: Surgical resection for gastric adenocarcinoma is associated with the restaging of esophageal cancer after neoadjuvant therapy. significant post-operative morbidity and mortality. The aim of this study was to Material and methods: We included 88 esophageal cancer patients without assess the prognostic significance of sarcopenia in patients undergoing cura - distant metastases, who were resected after neoadjuvant treatment. PET/CT tive resection for gastric adenocarcinoma with respect to post-operative mor- with a diagnostic, contrast-enhanced CT (CE-CT) scan, was performed before bidity, disease-free and overall survival. resection and the radiological staging was compared to postoperative histo- Material and methods: A retrospective analysis of a cohort of consecutive pathology as the reference standard. Locoregional stage (T, N-stage) was re- patients who underwent surgical resection for gastric adenocarcinoma be- trieved from the CT scan, whereas PET analysis consisted of measurement of tween 2008 and 2014 was conducted. Patient demographics, radiological pa- standardized uptake value (SUV) mean, SUV max, Tumor diameter, metabolic rameters and pathological data were collected. Osirix was used to measure tumor volume (MTV) and total lesion glycolysis (TLG). Sensitivity, specificity skeletal muscle area at a set level, which was normalized for height to calcu- and accuracy of the CT staging in reference to pathology were calculated. PET late skeletal muscle index. Sarcopenia entailed a skeletal muscle index of less 2 2 2 2 parameters were compared using analysis of variance (ANOVA). than 52.4cm /m for men and 38.5cm /m for women. Statistical analysis was Results: Of the 88 patients amenable for analysis, eleven patients (13%) had performed using the statistical package for the social sciences (SPSS). Re- received combination radiochemotherapy, whereas 75 patients (85%) had gression analysis was used to identify predictors of outcomes. neoadjuvant chemotherapy either in combination or monotherapy with various Results: Fifty-six patients (41 male, 15 female, mean age 68.4±11.9 years) met agents. For overall T-staging, true positive, false positive and false negative the inclusion criteria. Thirty-six percent (20/56) of patients were sarcopenic rates for CT vs. PET/CT (qualitatively) were: 59.1% vs. 68.1%, 25% vs 15.9% pre-operatively. Both sarcopenic and non-sarcopenic patient groups were and 15.9% vs. 15.9%, leading to an accuracy of 57.6% vs. 68.2%. For T3+4 equally matched with the exception of weight and body mass index (p=0.036 stages alone, CT had a better diagnostic performance (sensitivity 85.7, speci- and p=0.001). Sarcopenia was associated with a decreased overall survival ficity 81.8%). Volumetric PET parameters, but neither SUVmax nor SUVmean (Log Rank=0.003) and was an adverse prognostic predictor of overall survival correlated with T- stage: MTV (p=.011), TLG (p=0.35), PET-Diameter (p=0.001). on multivariable analysis (hazard ratio=10.915, p=0.001). Sarcopenia was a Likewise, there was an association to American joint committee on cancer predictor of in-hospital serious complications on multivariate analysis (odds (AJCC) stage: MTV (p=.015), TLG (p=0.37), PET-Diameter (p=0.002). ratio=3.508, p=0.042). Conclusion: In restaging of esophageal cancer after neoadjuvant therapy, Conclusion: In patients undergoing curative resection for gastric cancer, there both CE-CT and PET provide independent information on the post-therapeutic is a statistically significant association between sarcopenia and both de - locoregional tumor stage. creased overall survival and serious post-operative complications. The meas- urement and reporting of skeletal muscle index on pre-operative CT should be considered for patient preparation purposes. SS 5.5 More frequent follow-up CT examinations after surgical resection than endoscopic resection of early gastric SS 5.7 cancers: are they really useful? Prediction of tumor response to chemotherapy using S.E. Jung, M.H. Choi; Seoul/KR perfusion CT in patients with unresectable advanced gastric cancer Purpose: A recommended interval to follow-up CT or endoscopy after treat- 1 1 2 1 2 D.H. Lee , S.H. Kim , S.M. Lee ; Seoul/KR, AnYang/KR ment is not provided in the national comprehensive cancer network (NCCN) guidelines. No study has analyzed the role and interval of CT after both surgi- Purpose: To evaluate whether parameters obtained from perfusion CT (PCT) cal and endoscopic resection for gastric cancer. To evaluate the utility and can predict treatment response after palliative chemotherapy in patients with timing of CT in the detection of recurrent tumor after surgical and endoscopic unresectable advanced gastric cancer (AGC). resection (endoscopic submucosal dissection) for early gastric cancer (EGC). Material and methods: This prospective study was approved by the institu- Material and methods: A total of 670 patients after surgical (n = 535) and tional review board and informed consents were obtained. Twenty-one pa- endoscopic resection (n = 135) for EGC between 2007 and 2009 were enrolled. tients with unresectable AGC (M:F=13:8; mean age, 53.7 years old) were in- Mean numbers of CT and endoscopy between both treatment groups were cluded. We performed PCT and palliative chemotherapy in all 21 patients. compared. The mean and cumulative dose length product (DLP) of CTs were After palliative chemotherapy, treatment response was assessed for each pa- calculated. Recurrence-free survival and overall survival were evaluated using tient using RECIST ver. 1.1 and patients who achieved complete or partial re- a Kaplan-Meier method. sponse were classified into responder group. Relationship between tumor re - Results: The mean interval of CT was significantly shorter and the mean num - sponse to palliative chemotherapy and PCT parameters was evaluated using ber of CTs was significantly larger in the surgical resection group than in the the Mann-Whitney test and receiver operating characteristic analysis. endoscopic resection group. All 34 gastric recurrences were diagnosed by Results: After chemotherapy, 6 patients showing partial response were classi- endoscopy. Six extragastric recurrences were detected by CT out of total 5417 fied into the responder group and the remaining 15 patients into the non-re - CT examinations. The average cumulative DLP was significantly higher in the sponder group. Permeability surface value was significantly different between surgical than in the endoscopic resection group (P = 0.004). two groups (51.0 ml/100g/min in responder group versus 23.4 ml/100g/min in Conclusion: Follow-up CT was more frequently performed in patients after non-responder group, P=0.002). Other PCT parameters were not significantly surgical resection than in those after endoscopic resection of EGCs even different between two groups. The area under the curve to predict responders though CT showed limited usefulness for detecting recurrent tumors after ei- was 0.911 (95% confidence interval, 0.787-0.990; P=0.004) for permeability ther treatment methods. Since there was no extragastric recurrence in patients surface value and the sensitivity of 100% (6/6) and specificity of 80% (12/15) with mucosal cancer after resection, routine use of CT should be avoided after were achieved when a cut-off permeability surface value was set at 29.7 treatment of mucosal cancer. ml/100g/min. Conclusion: Obtaining perfusion parameters from PCT was feasible for pa- tients with unresectable AGC and can aid the prediction of treatment response after palliative chemotherapy as the permeability surface value in patients within the responder group was significantly higher than within the non-re - sponder group. S670 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / WEDNESDAY, JUNE 13, 2018 SS 5.8 SS 5.10 MRI of the gastric antrum for the quantification of Deep infiltrating endometriosis: can magnetic gastric motility: comparison between obese and normal resonance imaging predict the need for a colorectal weight patients surgeon? 1 1 2 3 3 S. Picchia , M. Rengo , M.A. Bali , D. Bellini , S. Badia , A. Brusic, S. Esler, L. Churilov, P. Chowdry, M. Sleeman, 1 1 2 3 A. Laghi ; Rome/IT, London/UK, Latina/IT P. Maher, N. Yang; Heidelberg, VIC/AU Purpose: To compare gastric motility in obese patients and normal weight Purpose: To identify Magnetic Resonance (MR) features that predict the need volunteers with magnetic resonance (MR). for colorectal (CR) surgeon intervention during gynaecologist-led laparoscopic Material and methods: This is a non-randomized single-centre study. The surgery for deep infiltrating endometriosis (DIE). motility analysis was performed on 1.5T MRI using 2DtrueFISP sequences on Material and methods: Retrospective cohort multivariate analysis. The cohort coronal and axial plane. Images were acquired before, immediately after a liq- consisted of women with laparoscopically-proven DIE, preoperative pelvic MR uid meal (Nutridrink plus 2 yolks, 1 albumen and 150 ml of water) and every 20 at Austin Health and subsequent surgery and histopathological confirmation at minutes for a total of 5 different acquisition times (T0-T5) and a total exam time Mercy Hospital (2006-2015, n=122). MRIs were reviewed independently by of 100 minutes. This 525-kcal meal was composed of 25% fat, 25% protein two radiologists blinded to outcomes. MR characteristics of pelvic endome- and 50% carbohydrate. Each gastric motility scan lasted 60 seconds, with 100 triosis were correlated to outcomes, to identify features associated with CR- images acquired in free breathing. Antral length (AL), maximal contraction am- surgeon intervention. plitude (MCA) and contraction frequency (CF) were evaluated on a dedicated Results: The readers demonstrated concordant findings for: presence of bow - software. Results obtained in obese patients and normal weight volunteers el lesions (kappa=0.68, 95%CI:0.55-0.81), bowel lesion length (Lin’s concord- were compared. ance coefficient 0.73, 95%CI:0.66-0.81); and bowel lesion depth (kappa=0.82, Results: Our final population consisted of twenty patients (10 obese and 10 95%CI:0.74-0.88). MR detected bowel lesions appeared more strongly associ- normal weight). MR showed that during fasting and postprandial period, in ated with the need for CR-surgeon intervention than current practice (defined obese patients all the evaluated parameters (AL, MCA and CF) were signifi - as the presence of a CR-surgeon in theatre), but this was not statistically sig- cantly lower than normal weight volunteers (mean basal AL obese nificant (area under receiver operating characteristic curve (ROC-AUC) 0.79 vs patients=6.75±0.8cm/normal weight patients=8.12±0.43cm, p value<0.05; 0.74, p=0.23). MR bowel lesions measuring ≥20mm or bowel lesions invading mean postprandial AL obese patients=7.56±1.03cm/normal weight the muscularis or submucosa, however, did demonstrate a statistically signifi - patients=9.4±0.12mm, p value<0.05; mean basal MCA obese cant correlation with CR-surgeon intervention and performed better than cur- patients=7.1±0.72mm/normal weight patients=8.97±0.43, p value<0.05; mean rent practice (ROC-AUC 0.84, p=0.014 and 0.90, p<0.0001, respectively). postprandial MCA obese patients=8.3±1.5mm/normal weight Conclusion: The current decision-making process to identify those patients patients=9.76±0.76mm, p value<0.05; mean basal CF obese with DIE that require a CR-surgeon to aid bowel resection is poorly defined and patients=8.7±0.98cpm/normal weight patients=10.3±0.9, p value<0.05; mean variable, resulting in sub-optimal resource allocation. This study identifies re - postprandial CF in obese patients=9.8±1.4cpm/in normal weight producible, objective MR features that better predict the need for CR-surgical patients=11.2±1.2cpm, p value<0.05). intervention than the current practice. This provides a basis for a prospective Conclusion: MRI identified differences in antral motility between obese pa - study for validation and development of a predictive scoring system to aid tients and normal weight volunteers. The obesity-related cause could become surgical planning and resource allocation. a potential new therapeutic target. SS 5.9 Utility of diffusion-weighted MRI with different b values in the evaluation of perianal fistulas M.H. Turkcanoglu, A.H. Yardımcı, C.T. Bektas, B. Kocak, O. Kilickesmez; Istanbul/TR Purpose: To determine the optimal b-value for diffusion-weighted imaging (DWI) in the diagnosis of perianal fistula and to demonstrate the diagnostic efficacy of this b-values and fistula tract thickness on DWI. Material and methods: Magnetic resonance (MR) images of 200 patients with the suspicion of perianal fistula referred to our Radiology Clinic were evaluated retrospectively. In our study, primary fistula tracts, internal orifices, abscess formations, horseshoe components and secondary tracts were evaluated. Axial fat-suppressed T2-weighted spin-echo (SE) images, DWI1 and DWI2 (b values, 400 and 1000s/mm ) contrast-enhanced T1-weighted images were analyzed for each patient by using a four-point scale. The combination of con- trast-enhanced T1-weighted images with T2-weighted images was used as reference standard. Results: 216 fistulas were detected in 200 patients (39 female,161 male). The sensitivity of DWI obtained at low b-value is higher in the evaluation of primary tract, internal orifice, secondary tract and horseshoe components in perianal fistulas. For the evaluation of abscess formations, it is seen that both low and high b-value DWI are equally sensitive. It has been shown that primary tract thickness is significant in detection of fistulas on DWI. The cut-off value for fistula thickness is 3.0 mm. Area under the curve 0.876 (0.831-0.922). Optimal cut-off value was obtained with 75.3% sensitivity, 100% positive predictive value, 100% specificity, 39.5% negative predictive value. Conclusion: The choice of optimal b values significantly affects the detection of perianal fistulas in DWI. Low b values (b=400s/mm(2)) had the best specific - ity and sensitivity. It has been shown that primary tract thickness is significant in detection of fistulas on DWI. S671 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 The Auditorium SS 6.3 Development and validation of a simplified magnetic Scientific Session SS 6 resonance index of activity for Crohn’s disease Small Bowel: imaging Crohn‘s disease J. Rimola, I. Ordas, S. Rodriguez, E. Ricart, J. Panés; Barcelona/ES SS 6.1 Purpose: To develop and validate a simplified and accurate magnetic reso - Diagnostic accuracy of layered pattern of enhancement nance index of activity (sMaRIA) for assessing activity and therapeutic re- during MRI exams in predicting active inflammation in sponse on patients with luminal Crohn’s disease (CD). Crohn’s disease: systematic review and meta-analysis Material and methods: Magnetic resonance enterography (MRE) data from 1 1 1 1 F. Rivosecchi , D. Bellini , D. Caruso , D. De Santis , 98 patients, including active and inactive segments (colon and terminal ileum), 1 2 1 2 M. Zerunian , A. Laghi ; Latina/IT, Rome/IT from two prospective studies were re-analysed to develop the sMaRIA using endoscopy (Crohn‘s disease endoscopic index of severity (CDEIS)) as the gold Purpose: To assess diagnostic accuracy of layered pattern (LP) on contrast- standard. Further analysis of responsiveness in a cohort of 37 patients who enhanced magnetic resonance imaging (MRI) in detecting active inflammation underwent MRE and endoscopy before and after 12 weeks of therapeutic in- in patients with Crohn’s disease. tervention was performed. Comparison between MaRIA and sMARIA for de- Material and methods: Methods for analysis were based on preferred report- tecting active/severe lesions and therapeutic response was performed. ing items for systematic reviews and meta-analyses (PRISMA). We searched Results: Logistic regression analysis showed that wall thickness >3mm, pres- MEDLINE, Cochrane library, Embase and Web of Science databases. Pooled ence of edema, ulcers and fat stranding were independent predictors of dis- estimates for sensitivity, specificity, pooled positive likelihood ratios (LR) and ease activity and were used therefore as descriptors of sMaRIA. The sensitiv- pooled negative LR were calculated using random effect model. I was used to ity and specificity of sMaRIA at segment level for detecting active disease evaluate heterogeneity. Quality assessment of diagnostic accuracy studies using a cutoff ≥5 were 90% and 81% (area under curve (AUC)=0.91, 95%CI (QUADAS) II was used to assess the quality of studies included. 0.88-0.94), and for detecting severe lesions (ulcers) using a cutoff ≥10 were Results: Of the 638 articles initially identified thorough databases searching, 7 85% and 92% (AUC=0.94, 95%CI 0.91-0.96). Correlation between sMaRIA were included for quantitative synthesis. Cumulative data for diagnostic ac- and CDEIS/MaRIA was excellent (r=0.84 and r=0.97, respectively; p<0.001). curacy on per patients basis were sensitivity 51% (95% CI: 44–59%; I : The sMaRIA accurately detected changes in lesion severity in response to a 78.5%), specificity 80% (95% CI: 74–85%; I : 91.4%), pooled positive LR 2.5 therapeutic intervention and was as reliable as endoscopy for the assessment 2 2 (1.32 to 4.73; I : 63.2%), pooled negative LR 0.67 (0.58 to 0.77; I : 0.0%). Het- of mucosal healing. erogeneity was high among studies. Conclusion: Simplified MaRIA index allows a faster and easier assessment of Conclusion: In conclusion, diagnostic accuracy of layered pattern on MRI has inflammation in CD by keeping high accuracy for both diagnosis and therapeu - a low sensitivity and a high specificity in detecting active inflammation in tic response. Main advantages over MaRIA includes, a less time consuming Crohn’s disease. Our results are limited due to the high heterogeneity among calculation and is not confounded by missing segments. studies included. SS 6.4 SS 6.2 Evaluation of image findings on CT enterography for Intra- and inter-observer variability in visually graded endoscopic complete remission after anti-tumor small bowel motility features from dynamic MRI of necrosis factor-alpha therapy in patients with Crohn’s Crohn’s disease patients 1 1 1 2 3 disease R. Gollifer , A. Menys , A. Plumb , F. Vos , J. Stoker , 1 1 1 2 S.H. Kim, J.E. Kim, T.O. Kim, J.-H. Yoon; Busan/KR D. Atkinson , S.A. Taylor ; London/UK, Delft/NL, Amsterdam/NL Purpose: To evaluate the image findings of CT enterography (CTE) obtained in patients with endoscopic complete remission (CR) after anti-tumor necrosis Purpose: Automated measurement of small bowel motility metrics is highly factor (TNF)-alpha therapy for Crohn’s disease. reproducible and is giving insights into the aetiology of abdominal symptoms Material and methods: Between August 2010 and October 2017, 36 con- in Crohn’s disease. It is unknown if a simple qualitative assessment of motility secutive patients with Crohn’s disease who received anti-TNF-alpha therapy by abdominal radiologists is sufficiently robust, and in particular is reproduci - and underwent pre-and post-therapy CTE as well as ileocolonoscopy were ble between observers. initially enrolled. CTE was performed with standard-dose enteric phase scan. Material and methods: 105 Crohn’s disease subjects underwent a 20-second Eleven patients were excluded due to a mismatch of follow-up intervals or a breath hold cine motility sequence, after mannitol ingestion, as part of mag- long interval (over 2 weeks) between CTE and ileocolonoscopy. Therefore, 25 netic resonance enterography. Five global small bowel motility metrics were patients (16 men, 9 women, mean age 26 years; range 18-43 years) were fi - visually graded by two experienced radiologists using a 10-point scale to as- nally analyzed. Two blinded readers reviewed pre-and post-therapy CTE im- sess 1) mean, 2) spatial variability, 3) temporal variability, 4) area of motile age findings in consensus as for active inflammation, i.e., mural hyperen - bowel and 5) distension quality. 26 datasets were analysed twice with a mini- hancement, mural thickening (thickness>3mm), mural stratifi cation, and mum gap of 2 weeks. Inter-and intra-reader variability was assessed using increased pericolonic fat attenuation in the rectum, colon, and terminal ileum, Bland-Altman plots. respectively. The endoscopic CR indicated mucosal healing identified by ile - Results: Inter-reader variability was generally poor for all motility features ocolonoscopy and served as the reference standard. graded e.g. the Bland-Altman mean difference in the spatial variability metric Results: Eleven patients had endoscopic CR. Six patients of them (6/11, 55%) was -0.76 au and the Bland-Altman 95% limits of agreement were -5.8 to +4.3 had residual mild mural thickening and mild mural hyperenhancement on post- across a range of values from 0.35 to 8.45. For radiologist 2, the worst agree- therapy CTE compared to pre-therapy CTE. The other 5 patients did not have ment was for the spatial variability metric with a mean difference between the any post-therapy CTE abnormalities. Non-endoscopic CR group (4 partial re- original grading and the repeat grading of -0.11 au and the Bland-Altman 95% missions and 10 poor remissions) showed a higher concordance rate of 86% adjusted limits of agreement were -4.1 to 3.9 across a range of values from 0.7 (12/14) with post-therapy CTE than endoscopic CR group did (45%, P=0.0358). to 7.4. Conclusion: Residual mild mural thickening and mild mural hyperenhance- Conclusion: Simple visual assessment of small bowel has poor inter and intra- ment were seen on post-therapy CTE over half of the patients having endo- observer agreement for most motility metrics and cannot replace automated scopic CR after anti-TNF-alpha therapy for Crohn’s disease. software measurement. S672 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 6.5 SS 6.7 Established damage detected at MRE in patients with Mural diffusion restriction pattern of small bowel: a Crohn’s disease in complete endoscopic remission possible tool for grading inflammation in Crohn’s J. Rimola, I. Alfaro, D. Vas, J. Castro, E. Ricart, J. Panés; disease Barcelona/ES N. Faluhelyi, O. Farkas, P. Bogner; Pécs/HU Purpose: To identify the Magnetic Resonance Enterography (MRE) lesions Purpose: To identify different diffusion restriction (DR) patterns in the small that persist in patients with Crohn’s disease (CD) in endoscopic remission as bowel wall and determine their diagnostic use in magnetic resonance enterog- indicators of established damage, and to determine its relationship with pre- raphy (MRE) examination of patients with Crohn‘s disease (CD). treatment MRE lesions. Material and methods: MRE (1.5 T GE Optima 450w MR) examinations were Material and methods: Patients with CD that had been included in prospec- performed after oral administration of mannitol (2.5%) solution. Diffusion- tive studies on autologous heamatopoyetic stem-cell transplant and/or anti- weighted images (axial DWI with b0, b400 and b800) and T2 images of 30 tumor necrosis factor (TNF) drugs were evaluated for this study. Inclusion cri- consecutive patients with small bowel CD activity were retrospectively ana- teria were: presence of at least one segment with severe inflammatory lesions lysed. In 8 patients, intravenous contrast agent (Gadovist) was also adminis- detected at MRE and achievement of endoscopic remission after 1 year of tered and pre- and postcontrast T1 images were added to the analysis. De- treatment. scriptive statistics, analysis of variance (ANOVA), Kruskal-Wallis and Results: 73 intestinal segments (28 patients) with severe inflammatory lesions Mann-Whitney tests were performed. at baseline achieved endoscopic remission after 1 year of treatment. The prev- Results: In 14 cases, DR was homogenously present in the inflamed bowel alence of creeping fat and bowel mural fat deposition did not change in asso- wall. Only mucosal DR was seen in 6 patients and DR showed a layered pat- ciation with the achievement of endoscopic remission (p=0.34 and p=0.35 re- tern (both mucosal and serosal restriction with a central band of relatively spectively). Furthermore, luminal strictures persisted in 50% of segments with lower restriction) in 10 cases. Restriction patterns were corresponding to the initial strictures, and wall thickness remained >3mm in 29% of segments with contrast enhancement pattern of all additional contrast-enhanced cases. Pa- severe inflammatory lesions at baseline. Regression analysis showed that pre - tients with layered pattern had a significantly higher Clermont score (median: dictive lesions on pre-treatment MRE for established residual mural lesions 31.68, range: 3.87) than the rest of the patients (median: 21.12 range: 18.94, after achieving endoscopic remission were the presence of bowel fat deposi- p=0. 04746). The three groups proved to have a different mural thickness with tion (OR=48.3, p=0.001) and strictures at baseline (OR=15.96, p=0.004), the thickest wall in the “layered” group (mean: 8.7 mm, SD: 0.82, p<0.00001). whereas creeping fat was the only predictive factor for persistent extramural Conclusion: The well-known contrast enhancement patterns of bowels in ac- lesions (OR=35.8, p<0.001) and for either mural or extramural lesions tive CD (homogenous, mucosal and layered) are also identifiable as DR pat - (OR=16.25, p<0.001). terns on DWI-MRE. Layered restriction pattern can be a marker of more severe Conclusion: Up to 41% of bowel segments with initial severe inflammation in activity of small bowel inflammation in Crohn‘s patients. long-standing endoscopic remission after treatment had residual lesions at MRE. Presence of bowel fat deposition, strictures and creeping fat at pre- SS 6.8 treatment MRE may predict the development of established damage. The utility of diffusion-weighted imaging in the characterisation and assessment of activity of small SS 6.6 bowel and colonic Crohn’s disease on MR enterography Preoperative evaluation of small bowel complications in J.P. Walsh, D. Byrne, H. Fenlon, C. Cronin; Dublin/IE Crohn‘s disease: comparison of diffusion-weighted and Purpose: To demonstrate the utility of diffusion-weighted imaging (DWI) in contrast-enhanced MR imaging characterisation and assessment of activity of small bowel and colonic Crohn’s 1 2 1 1 3 M. Barat , C. Hoeffel , M. Bouquot , R. Dautry , M. Boudiaf , disease on MR enterography (MRE). 1 1 1 1 1 K. Pautrat , R. Kaci , C. Eveno , M. Pocard , P. Soyer , Material and methods: Of 145 patients who underwent MRE over an 18 1 1 2 3 A. Dohan ; Paris/FR, Reims/FR, Meudon la Foret/FR month period (July 2016 – December 2017), we retrospectively analysed imag- Purpose: To compare diffusion weighted-imaging (DW) magnetic resonance ing performed for 63 consecutive patients (39 female, 24 male) with biopsy- imaging (MRI) to contrast-enhanced (CE) MRI in the preoperative assessment proven Crohn’s disease. Appearances on T2 weighted, T1 contrast-enhanced of small bowel complications of Crohn‘s disease (CD) with MR enterography sequences and DWI, where available, were assessed. Using a semi-quantita- (MRE). tive technique, MRE activity scores were calculated for involved segments. Material and methods: Thirty-eight patients who underwent surgery for CD Results were compared with endoscopy and histologic findings, where avail - complications and preoperative MRE were included. MRE examinations were able. blindly analyzed independently by one junior and one senior abdominal radi- Results: Of 71 segments analysed, 58 demonstrated abnormality on contrast- ologist for the presence of fistula, stenosis and abscesses. During a first read - enhanced or DWI imaging (50 small bowel, 8 colon). Abnormal contrast en- ing session T2-weighted images (T2-WI), steady-state sequences (True-FISP) hancement was identified in 48 of 50 abnormal small bowel segments and 8 of and DW-MRI were reviewed (set-1) and during a separate distant session, T2- 8 colonic segments. Abnormal DWI signal was identified in 25 of 31 small WI, True-FISP and CE-MRI were reviewed (set-2). Performances of each read- bowel segments and 8 of 8 colonic segments. DWI was unavailable for 26 seg- er for the two sets were evaluated using imaging, intraoperative and patho- ments. Calculated sensitivity of contrast-enhanced MRE for active small bow- logical findings as the standard of reference. el inflammation was 96.0% [86.3-99.5%], while that of DWI was 80.7% [62.5- Results: Forty-eight fistulas, 43 stenosis and 11 abscesses were deemed pre - 92.6%]. Analysis of colonic activity was confounded by small sample size. sent using the standard of reference. For the senior radiologist, sensitivities for Conclusion: While there was a trend towards greater sensitivity for contrast- the detection of fistulas, stenosis and abscesses ranged from 80% to 100% enhanced MRE over DWI for detection of acute small bowel inflammation, this for set 1 and from 88% to 100% for set 2 and specificity ranged from 56% to did not reach the level of statistical significance. While an ideal MRE protocol 70% for set 1 and from 53% to 93% for set 2, with no significant differences would likely include both sequences, this study suggests that DWI may pro- between the different sets (P-values, 0.342-0.429). For the junior radiologist, vide an alternative to contrast-enhanced sequences, especially for longitudinal higher performances were obtained with set 2 (P-values, 0.001 - 0.007). assessment of disease activity. Conclusion: For a senior radiologist, DW-MRE has similar sensitivities than CE-MRE for the diagnostic of CD complications. For a junior radiologist, CE- MRE yields better results than DW-MRE. S673 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 6.9 Magnetic resonance enterography (MRE) and enteric ultrasound (US) in newly diagnosed or relapsing Crohn’s disease: patient acceptability, perceived burden and preferences S.A. Taylor, A. Miles, T. Metric Investigators; London/UK Purpose: Magnetic resonance enterography (MRE) and ultrasound (US) are widely used to image Crohn’s disease. The aim was to compare patient ac- ceptability and burden of MRE and US to each other, and to other enteric in- vestigations, and identify drivers of preferences. Material and methods: 159 patients (mean age 38, 94 female) prospectively recruited to a multicentre diagnostic accuracy study comparing MRE and US completed a detailed experience questionnaire pertaining to the burden and acceptability of small bowel investigations, including MRE and US. Data was compared t-tests, chi-square tests and McNemar tests as appropriate. Predic- tors of burden and patient preference were tested using linear and logistic re- gression. Results: US, MRE and colonoscopy were rated as very/fairly acceptable by 98.6%, 88.3% and 60.0% of patients respectively (p<0.001). MRE recovery time was significantly longer than US (p<0.001), but shorter than colonoscopy (p<0.001). Patients were less willing to undergo MRE again than US (90.7% vs. 98.5%, p=0.012, but more willing than for colonoscopy (74.7%, p=0.017). MRE resulted in greater burden than US (t=13.56, df=144, p<0.001), although overall burden scores were low. Younger age and significant levels of emo - tional distress were associated with MRE burden. A majority (80.0%) preferred US to MRE. Higher MRE discomfort predicted a preference for US but patients rated test accuracy as more important than scan discomfort. Conclusion: MRE and US are well tolerated, although MRE generates greater patient burden, longer recovery times, and is less preferred than US. Patients, however, place greater emphasis on diagnostic accuracy than scan burden when determining preferences. SS 6.10 Accuracy of emergency CT examination without bowel preparation in the assessment of disease activity and complications in ileocolonic Crohn’s disease patients compared to standard magnetic resonance enterography S.G. Drago, S. Lombardi, A. Casiraghi, C. Talei Franzesi, S. Sironi, D. Ippolito; Monza/IT Purpose: To compare the agreement between abdomino-pelvic CT performed with intravenous contrast material without oral contrast administration, in an urgent setting at the Emergency Department, in comparison to MR-enterogra- phy (MRE) in detecting inflammatory bowel disease findings and its complica - tions in Crohn’s disease (CD) patients. Material and methods: Forty-eight patients with known or suspicion of CD underwent abdomino-pelvic contrast-enhanced CT study on a 256-row MDCT (iCT, Philips Medical Systems), without any previous bowel distention. After- ward, the same group of patients underwent MR-Enterography on a 1.5 T (Philips, Achieva MR system) using oral administration of 1.5Lt of contrast agent and intravenous contrast. Two readers independently reviewed all the images of both techniques and analyzed the following cross-sectional find - ings: bowel wall thickness and enhancement, mesenteric lymph-nodes, vas- cular engorgement, fibro-fatty proliferation, fistulas and abscesses. Results: Two readers assessed 56 diseased bowel segments with both mo- dalities: small bowel, colon-sigmoid and rectum. Inter-observer agreement was higher for bowel obstruction (k=0.650 CT, k=0.671 MRE), its complica- tions (k=0.810 CT, k=1 MRE) and lymphoadenopathy (k=0.810 CT, k=0.690 MRE). Higher inter-reader agreement was achieved for CT in comparison with MRE for wall thickening (k=0.750 CT, k=0.610 MRE), fibro-fatty proliferation (k=0.780 CT, k=0.523 MRE) and vascular comb sign (k=0.610 CT, k=0.360 MRE). Conclusion: Abdominal contrast-enhanced CT without bowel distention of- fers high diagnostic value in the assessment of bowel wall involvement and its complications in CD and, in selected cases, immediate MR-Enterography ex- amination could be postponed and reserved for follow-up. S674 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 The Liffey A SS 7.3 Supersonic B-ratio mode and shear-wave elastography Scientific Session SS 7 performances in detection and graduation of steatosis Diffuse and chronic liver diseases: current and liver fibrosis in patients with liver transplantation M. Dubois, V. Brun, M. Rayar, M. Auger, L. Beuzit, assessment of liver fibrosis and steatosis E. Quehen, Y. Gandon, B. Turlin, P. Houssel-Debry, A. Paisant; Rennes/FR SS 7.1 Purpose: To evaluate the diagnostic performance of Supersonic B-ratio mode Doppler ultrasonography devices, including and shear-wave elastography (SWE) for the assessment of steatosis and liver elastography, allows accurate screening for severe fibrosis in patients with liver transplantation. fibrosis Material and methods: After institutional review board approval and written P. Bazeries, J. Delahaye, V. Cartier, J. Lebigot, J. Boursier, informed consent, patients hospitalized from June 2017 to December 2017 for C. Aubé; Angers/FR a systematic check-up after liver transplantation were prospectively included. Purpose: To evaluate the contribution of the elastography following a Mode B Each patient underwent the same day SWE, liver biopsy (LB) and biological ultrasonography examination in the screening of severe hepatic fibrosis exams. Steatosis was measured using B-Ratio mode between the liver and (METAVIR ≥ F3). kidney and liver stiffness with 3 SWE consecutive measures. LB was used as Material and methods: Five hundred and fourteen patients were prospec- gold standard, grading steatosis S0 (0%), S1 (1-10%), S2 (11-30%) or S3 tively included in a monocentric six years study. All patients underwent a liver (>30%), and grading liver fibrosis from F0 to F4 according to the Metavir score. biopsy to establish the fibrosis stage (METAVIR scoring). The signs of hepatic Results: Fifty patients were included. Mean B-ratio value was significantly dysmorphism and portal hypertension were studied in ultrasonography ex- higher in patients with steatosis (0.94±0.13 vs 1.34±0.44, p<0.001). Sensitivity amination with Doppler. Liver stiffness evaluation was performed using both and specificity were, respectively, 77.3% (0.56-0.90) and 80% (0.58-0.92) the acoustic radiation force impulse (ARFI) (Siemens) and 2D shear-wave elas- (area under the receiver operating characteristic (AUROC)=0.82) for S≥1 with a trography (SWE - Supersonic Imagine) techniques. B-ratio cut-off value of 1.02; 84.6% (0.56-0.97) and 85.7% (0.68-0.95) (AU- Results: Three ultrasound parameters were independent predictors of severe ROC=0.92) for S≥2 with a B-ratio cut-off value of 1.13; and 85.7% (0.46-0.99) fibrosis: liver surface irregularity, spleen length (≥110 mm) and demodulation of and 76.5% (0.60-0.88) (AUROC=0.80) for S≥3 with a B-ratio cut-off value of the hepatic veins waveform. The presence of at least one of these 3 parame- 1.20. Mean SWE value for patients without significant fibrosis (≤F1) was ters had 85.6% sensitivity and 36.1% specificity for the diagnosis of severe 14.1±9.4 vs 27.4±0.31 for patient with fibrosis. fibrosis. The use of ARFI (diagnostic cut off ≥1.59m/s) after the detection of Conclusion: Supersonic B-ratio mode is a non-invasive and accurate method one of this three signs dramatically increased the specificity at 80.8%, at a to detect and graduate steatosis in patients with liver transplantation. Fibrosis price of a slight decrease in sensitivity (73.7%). Same results were obtained cut-off values seem to be higher in transplanted liver than in normal liver. with SWE (diagnostic cut-off ≥9.5kPa) with 73.3% specificity and 81.5% sen - sitivity. Addition of an elastographic measurement allows obtaining a positive SS 7.4 predictive value of approximately 50% for the diagnosis of severe fibrosis, A comparative evaluation of modified DIXON, CT (LAI) which is an acceptable rate to refer patients to a hepatologist for a complete and US elastography, in quantification of liver fat, with assessment of chronic liver disease. MR spectroscopy as reference Conclusion: The screening of severe hepatic fibrosis using 3 simple ultra - R. Jain, S. Kale, N. Panchal, N. Kundaragi, sound signs with addition of elastographic measurement in case of positivity B. M, R. Talwade, N. Reddy; Bangalore/IN of one of them is a feasible and accurate procedure. Purpose: Early diagnosis and treatment of fatty liver disease, along with mod- ifications to lifestyle, helps reduce the rate of progression to end stage liver SS 7.2 disease and hepatocellular cancer. Additionally, liver fat is a risk for post-oper- Diagnostic performance of texture analysis for the ative complications of a transplant. Non-invasive MRI can be used to monitor assessment of hepatic fibrosis in patients with liver fat. This study was conducted to compare the accuracy of modified DIX - nonalcoholic fatty liver disease (NAFLD): preliminary ON, CT(liver attenuation index (LAI)) and US Elastography, with MR Spectros- results and comparison with MR elastography (MRE) 1 2 2 2 2 copy (MRS) as a reference benchmark in quantification of liver fat. R. Cannella , M. Tublin , A. Borhani , J. Behari , A. Furlan ; 1 2 Material and methods: Tests were performed on 80 patients (35 females and Palermo/IT, Pittsburgh, PA/US 45 males) which were grouped into hepatosteatosis-alcoholic (n=15), non-al- Purpose: To investigate the performance of texture analysis (TA) in the quanti- coholic non-obese (n=8), obese (n=23), liver donors (n=25) and post-chemo- fication of fibrosis in NAFLD and to compare it with MR elastography (MRE). therapy (n=9). MRI images were obtained on 3T Philips Ingenia, CT indices Material and methods: 54 adult patients (33 females, 21 males) with biopsy- calculated on Philips 128 Slice Ingenuity Core Scanner, US Elastography per- proven NAFLD were prospectively enrolled and underwent MRI on a 1.5T sys- formed on Philips Affinity70. Results were independently interpreted by two tem using a combination of T1-weighted gradient-recalled-echo (T1W 3D- radiologists. All data and interpretations were compiled and compared for pa- GRE) sequence (slice thickness =4.6 mm) and 2D-MRE. Texture analysis was tient groups, CT- LAI, elastography score (F0-F4), percentage of liver fat performed using a commercially available research software (TexRAD) on axial through mDIXON and MRS. T1W images. A circular region of interest (ROI) (3 cm ) was traced on an image Results: Sensitivity values are 97% for MR Spectroscopy, 96.5% for mDIXON, at the level of the porta hepatis. Receiver operating curves (ROC), areas under 71% for CT (LAI) and 64% for US elastography. The study showed excellent the ROC (AUROC) and 95% confidence intervals were calculated to assess correlation between MR spectroscopy and mDIXON (r value of 0.93), moder- the accuracy of each textural parameter and of liver stiffness for the diagnosis ate correlation with CT-LAI (r value of 0.89) and poor correlation with US elas- of significant fibrosis (SF: F≥2) and advanced fibrosis (AF: F≥3). AUROCs were tograpy (r value of 0.003). 15 out of 80 patients underwent biopsy, the histo- compared using the DeLong test. pathology results for these were included in the study. Results: Thirty-seven subjects had SF and 20 had AF. The textural parameters Conclusion: Modified DIXON is as accurate as MR spectroscopy as a nonin - with the best performance were SD and entropy with AUROC 0.755 (0.619- vasive method for quantification of hepatosteatosis, compared to the widely 0.862, p≤0.0002) and 0.769 (0.634-0.873, p<0.0001) for SF and AUROC 0.746 used CT (LAI) and US elastography which have their inherent drawbacks of (0.609-0.854, p≤0.0004) and 0.754 (0.618-0.861, p≤0.0002) for AF. The AU- radiation and operator dependence. ROC of MRE for SF was 0.785 (0.652–0.885, p≤0.0001) and for AF was 0.923 (0.817-0.978, p≤0.0001). While the AUROC of MRE for AF was significantly higher than that of SD and entropy (p=0.03 and p=0.04), there was no signifi - cant difference among AUROCs for SF (p=0.85 and p=0.74). Conclusion: Texture analysis has fair accuracy for the quantification of he - patic fibrosis in NAFLD, although the performance is inferior to MRE for the assessment of AF. S675 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 7.5 SS 7.7 Clinical value of ultrasonic adaptive sound speed Reliability criteria for liver stiffness measurement with estimation for the diagnosis and quantification of acoustic radiation force impulse 1 2 1 1 2 hepatic steatosis J. Boursier , C. Cassinotto , V. Cartier , A. Lannes , S. Shili , 1 2 1 1 1 2 1 2 M. Dioguardi Burgio , M. Imbault , M. Ronot , A. Faccinetto , J. Lebigot , B. Lapuyade , P. Cales , J.-B. Hiriart , 1 1 1 2 1 3 2 1 B. Van Beers , P.-E. Rautou , L. Castera , J.-L. Gennisson , S. Michalak , B. Le Bail , A. Mouries , F. Oberti , 2 1 1 2 1 2 2 1 M. Tanter , V. Vilgrain ; Clichy/FR, Paris/FR I. Fouchard , F. Chermak , V. De Ledinghen , C. Aubé ; 1 2 3 Angers/FR, Pessac/FR, Bordeaux/FR Purpose: To evaluate the ability of a new ultrasound (US) method based on Sound Speed Estimation (SSE) in detection, quantification, and grading of he - Purpose: To determine the reliability criteria of liver stiffness measurement patic steatosis using magnetic resonance (MR) proton density fat fraction (LSM) using acoustic radiation force impulse (ARFI) in chronic liver diseases. (PDFF) as a reference standard and to calculate one US fat index based on the Material and methods: 1094 patients with chronic liver diseases had liver bi- patient’s SSE. opsy and LSM with ARFI in two centers. Advanced fibrosis was defined as Material and methods: We consecutively included N=50 patients as study non-alcoholic steatohepatitis clinical research network (NASH CRN) histologi- cohort and further N=50 as validation cohort who underwent both SSE and cal scoring system F≥3 or Metavir F≥2. abdominal MR. Hepatic steatosis was classified according to MR-PDFF cut - Results: LSM failure rate was 0.9% (10/1094). Among the remaining 1084 offs as S0 ≤6.5%, S1 from 6.5 to 16.5%, S2 from 16.5 to 22% and S3≥ 22%. patients (male: 61.9%; age: 54.3±13.3 years), 48.5% had advanced fibrosis Receiver operating curve analysis was performed to evaluate the diagnostic and 16.8% cirrhosis. Etiologies of chronic liver disease were non-alcoholic performance of SSE in diagnosis of steatosis. Based on the optimal data fit fatty liver disease (NAFLD) (48.5%), viral hepatitis (26.3%), alcohol (12.2%), derived from our study, we proposed a correspondence between the MR- and other (13.0%). ARFI accuracy decreased with increasing interquartile PDFF and an US fat index. Coefficient of determination R was used to evalu- range (IQR) of liver stiffness to the median (IQR/M) ratio, especially for interme- ate fit quality and was considered as robust when R >0.6. diate/high levels of liver stiffness. Three reliability categories were thus de- Results: Study and validation cohort presented mean SSE value of fined: “very reliable” (IQR/M <0.15), “reliable” (0.15≤ IQR/M <0.35 or IQR/M 1.570±0.026 and 1.568±0.023 mm/µs for S0 and 1.521±0.031 and 1.514±0.019 ≥0.35 with ARFI median <1.37 m/s), and “poorly reliable” (IQR/M ≥0.35 with mm/µs for S1-S3 (p<0.01) patients, respectively. SSE threshold of ≤1.537 mm/ ARFI median ≥1.37 m/s). Using these criteria, the rates of patients correctly µs had 80% sensitivity and 85.7% specificity in diagnosis of steatosis in the classified were, respectively: 80.9%, 73.7%, and 57.8% for advanced fibrosis study cohort. Robust correspondence between MR-PDFF and the US fat in- (p=0.029 between very reliable and reliable; p <0.001 for other paired com- dex was found both for the study cohort (R = 0.73) and the validation cohort parisons); and 92.6%, 83.4%, and 50.0% for cirrhosis (p<0.001 for all com- (R = 0.76). parisons). 23.6% of the ARFI examinations were very reliable, 55.0% reliable, Conclusion: SSE can be used to detect, quantify and grade liver steatosis and and 21.4% poorly reliable. The skin-liver capsula distance was an independent to calculate an US fat index. predictor of poorly reliable LSM, which occurred in 52.7% of patients having a distance ≥30mm. Conclusion: Based on IQR/M ratio and ARFI median, we defined three reliabil - SS 7.6 ity categories for LSM using ARFI, associated with different diagnostic accura- Transducer-free hepatic magnetic resonance cies. These new reliability criteria will increase confidence in ARFI results and elastography using cardiac wave induction at 0.3 ms diffusion of this technique. temporal resolution 1 2 1 1 M.A. Troelstra , J.H. Runge , A. Polcaro , O. Darwish , 1 3 2 1 J. De Arcos , T. Schneider , A.J. Nederveen , R. Sinkus ; SS 7.8 1 2 3 London/UK, Amsterdam/NL, Guildford/UK Non-invasive monitoring of hepatic steatosis via acoustic structure quantification of US with MR Purpose: This study aims to develop a transducer-free hepatic magnetic reso- spectroscopy as a reference standard nance elastography (MRE) method to facilitate clinical MRE implementation. D.H. Lee, J.Y. Lee; Seoul/KR We propose a cardiac-triggered motion-sensitised pencil-beam, for capturing cardiac-induced hepatic shear-waves at a high temporal resolution (>0.3 ms). Purpose: To prospectively evaluate whether monitoring hepatic steatosis by Material and methods: Six healthy volunteers were scanned on a Philips 3T- ultrasonography with an acoustic structure quantification (ASQ) technique is scanner. The pencil-beam navigator was positioned below the heart in liver feasible when using magnetic resonance spectroscopy (MRS) as a reference segments 2-4. ECG-triggered sequences obtained 35-40 pencil-beam images standard. (d=30mm, L=80mm, t=20ms) acquired in one R-R interval and repeated for 60 Material and methods: Thirty-six patients with suspected fatty liver disease successive R-R intervals, divided over four breath-holds, attaining temporal underwent both ultrasonography with ASQ and MRS on the same day. After a resolutions >0.3 ms. Phase-images yielded space-time images of hepatic mean follow-up period of 11.4±2.5 months, follow-up ultrasonography with transient shear-wave propagation. Straight lines were fitted along phase per - ASQ and MRS were performed on 27 patients to evaluate whether hepatic st nd turbations at 1 and 2 heart sound, to determine shear wave speed and shear steatosis improved. The focal disturbance (FD) ratio, as calculated using ASQ, stiffness μ=ρc (assuming no loss-effects). and the hepatic fat fraction (HFF), estimated by MRS, were obtained at both Results: Analysis showed two shear-wave propagations from the heart into initial and follow-up examinations. Pearson’s correlation coefficient was calcu - st nd the liver, at 1 (atrioventricular-valve closure) and 2 heart sounds (aortic-/ lated to assess correlations between ordinal values. nd pulmonary-valve closure). 2 heart sound waves showed an average wave Results: The FD ratio showed a strong, negative linear correlation with the speed of 1.49+/-0.07 m/s and stiffness of 2.24+/-0.21 kPa, 2x higher shear HFF after logarithmic transformation of both variables from the initial examina- nd stiffness compared to classical-MRE at 40Hz. The 2 heart sound shows a tions of 36 patients (ρ=-0.888; P<0.001) and the follow-up examinations of 27 frequency emission spectra peak at 85Hz. Liver-tissue properties are approxi- patients (ρ=-0.920; P<0.001). There was also a significant, negative linear cor - mately linear with frequency, resulting in stiffnesses of ~1kPa measured at relation between the change in the logarithm of the FD ratio and the change in 40Hz increasing to ~2kPa at 85Hz. the logarithm of the HFF by MRS over the follow-up period (ρ=-0.645; Conclusion: This approach allows transducer-free visualisation of hepatic P<0.001). In 16 patients with increased FD ratio on follow-up, HFF on follow- cardiac-induced shear-waves at high temporal resolution (>0.3ms), using up MRS significantly decreased, and HDL level significantly increased whereas waves created by cardiac-valve closure. Results show fast wave propagation, LDL tended to decrease. caused by high-frequency content of waves generated from cardiac-valve clo- Conclusion: The FD ratio was significantly correlated with the HFF at both the sure. This transducer-free technique could facilitate clinical MRE implementa- initial and follow-up examinations, and there was also a significant correlation tion, once it is validated in a larger population. between the changes in the FD ratio and the changes in the HFF over the fol- low-up period. S676 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 Liffey Hall 2 SS 7.9 Molecular features, histological parameters and Scientific Session SS 8 enhancement measurements in nonalcoholic steatohepatitis patients using gadoxetic acid-enhanced Malignant pancreatic tumours: new trends MR Imaging (preliminary results) 1 2 1 1 N. Bastati-Huber , D.S. Feier , A. Beer , S. Pötter-Lang , SS 8.1 1 1 1 H. Einspieler , A. Ba-Ssalamah ; Vienna/AT, Lower energy levels and iodine density images increase Cluj-Napoca/RO pancreatic tumor conspicuity on rapid kV-switching dual Purpose: To assess the relationship of organic anion transporting polypeptide energy CT (OATP1B1/3) expression with histological parameters and relative enhance- S. Aslan, M.S. Nural, İ. Camlidag; Samsun/TR ment ratio in patients with nonalcoholic fatty liver disease (NAFLD), defined Purpose: To evaluate the efficacy of rapid kV-switching dual energy CT (rs - according to the steatosis activity and fibrosis (SAF) scoring system, which is DECT) energy level optimization and iodine density (ID) images on pancreatic based on the semiquantitative scoring of steatosis activity and liver fibrosis. cancer detection and conspicuity. Material and methods: The local institutional review committee approved this Material and methods: 56 patients with pancreatic adenocancer were pro- study and waived written informed consent. This was a retrospective study of spectively enrolled and underwent rsDECT in pancreatic protocol. Largest le- gadoxetic acid-enhanced 3T MR imaging performed in 52 consecutive pa- sion diameters (LLDs) were measured on 45 keV, optimal CNR, 70 keV and ID tients with NAFLD (mean age (SD), 50.46 (16.52) years. The MR images were images. Attenuation differences between the lesion and normal parenchyma analyzed by using the relative enhancement (RLE) (the ratio of signal intensities (HU), image noise for each energy level, contrast gain and lesion contrast-to- of the liver parenchyma before and 20 minutes after intravenous administration noise ratio (CNR) were calculated. Iodine content of the lesions and the paren- of gadoxetic acid). Univariate and multiple regression analyses were applied to chyma were measured on ID images. Isodense lesions were treated as a sub- identify variables associated with OATP1B1/3 expression. group and all measurements were also applied to them. All analyses were Results: OATP1B1/3 expression correlated with RLE (r=0.49, p=0.0002), the performed by two radiologists at different times and inter-reader agreement degree of liver fibrosis (r=-0.27, p=0.04) but not with steatosis (r=-0.16, was evaluated. p=0.23), ballooning (r=-0.22, p=0.1) or inflammation (r=-0.07, p=0.57). Accord - Results: Optimal CNR was 52 ± 8,5 kEv. LLDs were measured on ID images ing to multivariate analysis both RLE and liver fibrosis are independent predic - (p< 0.001). Attenuation measurements on 45 keV and optimal keV were sig- tors of OATP1B1/3 expression (coefficient of determination R2=0.32, p=0.01). nificantly higher than 70 keV images. Attenuation difference between the lesion Conclusion: The degree of OATP1B1/3 expression correlated statistically with and parenchyma was significantly different between 70 keV images and other gadoxetic acid relative enhancement and the degree of liver fibrosis in patients energy levels but not different between 45 keV and optimal CNR. Highest im- with NASH. age noise was calculated in 45 keV. Highest contrast gain was between 70 keV and optimal CNR and highest CNR was measured in optimal CNR. Iodine con- SS 7.10 tent of tumors were significantly lower than the parenchyma (1,83 mg/cc vs The positive effects of bariatric surgery on non-alcoholic 3,05 mg/cc, p < 0.05). Isodense tumors also showed similar results. Inter-read- fatty liver disease: evaluating the changes in the liver fat er agreement was high (ICC 0,78 - 0,93). fraction and liver dimensions by using ideal IQ Conclusion: rsDECT is a reliable method for pancreatic cancer detection and sequences delineation, particularly isodense tumors. Y. Metin, N. Orhan Metin, O. Özdemir, M.E. Kadıoğlu, E. Beykoz Çetin, S. Kalcan, M.K. Çolakoğlu; Rize/TR SS 8.2 Purpose: To evaluate the early and intermediate effects of bariatric surgery on Prediction of residual tumor classification and overall liver fat fraction, liver length and liver volume using ideal IQ sequences at 3T survival in pancreatic cancer by preoperative CT 1 1 1 2 1 MRI. J.S. Bae , J.H. Kim , I. Joo , W. Chang , J.K. Han ; 1 2 Material and methods: Body mass index (BMI), body weight, liver fat fraction Seoul/KR, Seongnam/KR (L-FF), liver volume and craniocaudal length were measured preoperatively (1 Purpose: To predict resection (R) stage classification and overall survival on week before surgery) and postoperatively (1 and 6 months after surgery) in 46 preoperative CT in patients who underwent surgery for pancreatic cancer. patients (31 female, 15 male; age range, 21–60 years) who underwent bariatric Material and methods: In this retrospective study, 216 patients with pancre- surgery between May 2016 and August 2017. Liver volumes, L-FF were calcu- atic cancer who underwent CT and surgery were included. Preoperative CT lated from Ideal IQ sequences. findings were assessed by two radiologists and R classification was catego - Results: BMI decreased from 46.3 ± 5.5 to 40.7 ± 5.3 (at first month), and to rized into no residual tumor (R0) and residual tumor (R1 or R2). We assessed 32.6 ± 5.5 kg/m at the end of sixth month. Mean liver volume decreased from the correlation between CT findings and R classification. In addition, we used 2641.2 ± 543.6 to 1954.1 ± 384.2 cm (at first month) and to 1776.3 ± 361.1 Kaplan-Meier estimation for survival analysis and Cox proportional hazard cm at the end of sixth month. L-FF decreased from 16.4 ± 10.2 to 7.2 ± 4.9% model to find prognostic factors for overall survival. (at first month) and to 4.0 ±2.7% at the end of sixth month. At the first month, Results: There were 153 patients in R0 group and 63 patients in R1 or R2 60.9% of patients and at the sixth month 82.6% of patients had resolution of group. Tumor size (odds ratio (OR) 1.045, 95% confidence interval [CI]: 1.005- steatosis. 1.086), definite peritumoral fat stranding (OR 3.826, 95% CI: 1.765-8.293), Conclusion: A significant decrease in liver volume, liver length and FF was portal vein involvement (OR 2.795, 95% CI: 1.216-6.423), and suspicious dis- achieved at first and sixth months after bariatric surgery. The reduction in L-FF tant metastasis (OR 2.916, 95% CI: 1.296-6.561) were independent predictors is most prominent after the first month and continues at sixth month. The pos - for residual tumor (P<0.05). On survival analysis, median survival period was itive effects of bariatric surgery on liver dimensions and FF can be monitorized 29.2 months in R0 group and 12.2 months in R1 or R2 group. T4 stage (hazard by using ideal IQ sequences. ratio (HR) 4.330, 95% CI: 1.208-2.555), N1 stage (HR 1.757, 95% CI: 1.208- 2.555), common hepatic artery involvement (HR 4.821, 95% CI: 2.125-10.934) and R stage (HR 2.499, 95% CI: 1.649-3.789) were predictors of poor survival (P<0.05). Conclusion: Preoperative CT is useful to predict R classification using tumor size, definite peritumoral fat stranding, portal vein involvement, and suspicious distant metastasis, as well as to anticipate poor survival using T stage, N stage, common hepatic artery involvement and R classification. S677 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 8.3 SS 8.5 Radiological assessment of local resectability status in Perfusion measurements in pancreatic cancer by means patients with pancreatic cancer: interreader agreement of intravoxel incoherent motion MRI and CT perfusion: and reader performance at two different classification correlation with each other and with histological systems microvessel density 1 1 1 2 N. Kartalis , C. Valls , E. Axelsson , M. Andersson , P.F.J. Mayer, F. Fritz, S. Skornitzke, W. Stiller, M. Klauss, 3 4 5 1 I. Keussen , J. Strinnholm , W. Bartholomä , M. Del Chiaro , H.U. Kauczor; Heidelberg/DE 1 1 1 1 R. Segersvärd , L. Lundell , L. Loizou ; Stockholm/SE, Purpose: The aim of this study was to compare intravoxel incoherent motion 2 3 4 5 Gothenburg/SE, Lund/SE, Umeå/SE, Linköping/SE (IVIM) diffusion-weighted (DW) MRI and CT perfusion to assess tumor perfu- Purpose: To assess the interreader agreement and reader performance in the sion of pancreatic ductal adenocarcinoma (PDAC). evaluation of patients with pancreatic cancer (PC) at two classification sys - Material and methods: In 19 patients with PDAC, DW MRI and CT perfusion tems of local resectability status, namely the national comprehensive cancer were conducted on the day before surgery. IVIM analysis of DW MRI was per- network (NCCN) and Karolinska Institutet’s classification system (KICS). formed and the parameters perfusion fraction f, pseudodiffusion coefficient D*, Material and methods: In this institutional review board-approved retrospec- and diffusion coefficient D were extracted. The CT perfusion parameters blood tive study, six radiologists evaluated independently pancreatic CT-examina- volume (BV) and blood flow (BF) were estimated with a deconvolution-based tions of 30 patients randomly selected from a tertiary referral centre’s multidis- analysis. In 10 patients, intratumoral microvessel density (MVD) and microves- ciplinary tumour board database. Based on well-defined criteria of sel area (MVA) were analyzed from representative immunostained tissue slides, tumour-vessel relationship, each patient was assigned into one of three NCCN using an anti-CD34 antibody as endothelial marker. Pearson correlation coef- and six KICS categories. We assessed the intraclass correlation coefficient ficients between these parameters were calculated. (ICC; a score of 0–0.2 indicates poor agreement, 0.3–0.4 fair agreement, 0.5– Results: There were significant positive correlations between MVD/ MVA and 0.6 moderate agreement, 0.7–0.8 strong agreement, and >0.8 a very strong f/ BF (r ≥ 0.666, p ≤ 0.036). f significantly positively correlated with BF and BV agreement) and compared the percentages of correct tumour classification of (r ≥ 0.691, p ≤ 0.001). Correlation coefficients between BF and MVD/MVA were the six readers at both systems (Chi-square test; a P-value <0.05 was consid- not significantly different from correlation coefficients between f and MVD/ ered significant). Standard of reference was a consensus evaluation of CT from MVA (p ≥ 0.796). three readers not involved in the analysis. Conclusion: The study shows that, in PDAC, CT perfusion derived BF and Results: The ICC for NCCN and KICS was 0.82 and 0.84, respectively (very IVIM derived f similarly reflect microvascularity and could possibly serve as strong agreement). The percentages of correct tumour classification at NCCN imaging biomarkers for tumor characterization. and KICS were 53-83% and 27-57%, respectively, with no statistically signifi - cant differences in the overall reader comparison per classification system. SS 8.6 Conclusion: Interreader agreement at both PC classification systems is very Relationship between pathologic treatment response strong. NCCN may be advantageous in terms of reader performance com- after neoadjuvant therapy and metabolic response on pared to KICS. integrated time-of-flight fludeoxyglucose (FDG) positron-emission tomography/MRI in patients with SS 8.4 FDG-avid borderline resectable pancreas cancer: a Intravoxel incoherent motion diffusion-weighted MR feasibility study imaging: reproducibility and diagnostic value for N. Samreen, I. Garg, J.G. Fletcher, M. Truty, G.B. Johnson, characterization of solid pancreatic lesions J.L. Fidler, K. Bradley, A. Goenka; Rochester, MN/US 1 2 1 R. De Robertis , N. Cardobi , P. Tinazzi Martini , Purpose: To determine if fludeoxyglucose (FDG) positron-emission tomogra - 3 4 4 1 M. Zanirato , A. Stemmer , R. Grimm , M. D‘Onofrio ; phy (PET)/MRI can be used to predict pathologic treatment response after 1 2 3 4 Verona/IT, Peschiera del Garda/IT, Milan/IT, Erlangen/DE neoadjuvant therapy in FDG-avid borderline resectable pancreatic ductal ad- Purpose: To evaluate the reproducibility and the diagnostic potential of intra- enocarcinoma (PDAC). voxel incoherent motion (IVIM)-derived parameters for differentiation of solid Material and methods: Patients with FDG-avid PDAC on baseline integrated pancreatic lesions and normal pancreas. time-of-flight PET/MRI who also underwent a post-neoadjuvant therapy PET/ Material and methods: Forty-seven patients with solid pancreatic tumors (30 MRI prior to surgical resection were included. Primary tumor SUVmax, SU- ductal adenocarcinomas (PDACs), and 17 neuroendocrine neoplasms (panN- Vmean and volumetric PET parameters on pre- and post-neoadjuvant therapy ENs)), 5 patients with mass-forming pancreatitis (MFP), and 30 subjects with scans were measured using anatomic guidance from simultaneously acquired normal pancreas (NP) were included. All subjects underwent 1.5 T MR imaging contrast-enhanced MRI. Metabolic response on PET/MRI was correlated to including IVIM diffusion-weighted imaging with 11 b values (from 0 to 800 sec/ histologic treatment response using College of American Pathologists grading mm ). Apparent diffusion coefficient (ADC), true diffusion (D), pseudodiffusion system (path grade). Complete metabolic response was defined as FDG up - (D*), and perfusion fraction (f) were calculated by two independent readers. take indistinguishable from surrounding background and normalization of Interobserver reliability of the measurements was assessed by using the intra- post-therapy CA 19-9 were evaluated as surrogates of path grade 1/0. class correlation coefficient (ICC). A Kruskal-Wallis H test was used for com - Results: 12 patients (range 55-79years; 50% males) underwent required PET/ parison. The diagnostic performance of each parameter was evaluated using MRI, neoadjuvant therapy, and surgical resection. Treatment response grades receiver operating characteristic (ROC) analysis. were none (grade 3, n=2), moderate (grade 2, n=6), marked (grade 1, n=2) and Results: Interobserver agreement was excellent for pancreatic lesions (ICC = complete pathologic response (pCR) (grade 0, n=2). There was no significant .981, .985, .963, and .989 for ADC, D, D* and f, respectively). ADC was not difference (p>0.05) in baseline PET parameters between group 1 (grade 3/2) significantly different between lesions. F was significantly higher in NP and and group 2 (grade 1/0). Complete metabolic response on post-therapy PET/ panNENs compared with PDACs (both p<.001). D was significantly higher in MRI was observed in 5 patients – one with grade 2, and two each with grades PDACs compared with NP and panNENs (p<.001 and .001). D* was signifi - 1 and 0. Positive and negative predictive values of complete metabolic re- cantly lower in MFP compared with panNENs, in PDACs compared with NP, sponse for path grade 1/0 were 80% and 100% whereas for normalization of and in PDACs compared with panNENs (p=.047, .047, and <.001, respective- post-therapy CA 19-9 were 20% and 33%, respectively. ly). For the differentiation between PDACs and panNENs, f had the highest Conclusion: FDG PET/MRI shows promise for response evaluation following area under the curve (.976) in ROC analysis. neoadjuvant therapy with complete metabolic response by PET/MRI correlat- Conclusion: The measurement of IVIM-derived parameters in pancreatic le- ing with marked or complete pathologic response. sions has excellent reproducibility. IVIM-related parameters could be helpful in distinguishing PDACs from panNENs. S678 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 8.7 SS 8.9 MRI imaging of IPMN: evaluation of agreement between Percutaneous radiofrequency ablation of pancreatic observers with different degrees of experience adenocarcinoma 1 2 2 2 1 1 2 1 L. Bertuzzo , G.A. Zamboni , G. Cardano , R. Pozzi Mucelli , A. Sarno , G. Tedesco , R. De Robertis , S. Paiella , 2 1 2 1 2 2 1 1 G. Mansueto ; Santorso-VI/IT, Verona/IT R. Salvia , I. Frigerio , R. Girelli , D. Melisi , C. Bassi , 1 1 2 M. D‘Onofrio ; Verona/IT, Peschiera del Garda/IT Purpose: The Sendai and Fukuoka consensus criteria were established to as- sess the risk of malignancy in pancreatic mucinous cystic lesions. The purpose Purpose: The objective of this study is to evaluate the feasibility and safety of of this study was to assess the agreement between readers with different ex- percutaneous radiofrequency ablation (RFA) of locally advanced pancreatic perience in recognizing malignancy features in a series of IPMN. cancer located in the pancreatic body. Material and methods: Institutional review board (IRB) approval was waived Material and methods: Patients with biopsy-proven locally advanced pancre- for this retrospective study. We included 118 patients (45 M, 73 F, average age atic adenocarcinoma were considered for percutaneous radiofrequency abla- 68 years) with a diagnosis of IPMN who underwent MRI/MRCP in our center. tion. Postprocedural CT studies and CA 19.9 tumor marker evaluation were Two readers, respectively with 10 and 4 years of experience in abdominal im- performed at 24 hours and 1 month. At CT, the treatment effect was evaluated aging, reviewed independently the scans assessing for wall/internal septa by excluding the presence of complications. The technical success of the pro- thickening, mural nodules, dilation of the main pancreatic duct (MPD) and con- cedure is defined at CT as the achievement of tumoral ablated area. trast enhancement. The readers were blinded to clinical and laboratory infor- Results: Thirty-five patients have been included in the study. Five of the 35 mation. Kappa statistics was calculated. patients were excluded. At CT, the mean size of the intralesional postablation Results: 10 patients had mixed-type IPMN and 108 branch-duct IPMN. Pa- necrotic area was 32 mm (range: 15-65 mm). None of the patients developed tients had a median of 2 cysts >5 mm each (range 1-20; mean 2.6). A total of postprocedural complications. Mean CA 19.9 serum levels 1 day before, 1 day 307 cystic lesions were reviewed, with a mean size of 11.8 mm (5-50 mm). The after, and 1 month after the procedure were 285.8 U/mL (range: 16.6-942.0 U/ two readers showed very good agreement regarding MPD dilation (k=0.908), mL), 635.2 U/mL (range: 17.9-3368.0 U/mL), and 336.0 U/mL (range: 7.0- wall thickening (k=0.893), and presence of mural nodules (k=0.856). The 1400.0 U/mL), respectively. The mean survival after RFA procedure of the pa- agreement was good for presence of filling defects (k=0.721), wall enhance - tients, calculated on the data collected for 26 subjects, is 312 days (range: ment (k=0.725) and presence of mural nodules (k=0.663). 65-718 days). Conclusion: Readers with different levels of experience in abdominal imaging, Conclusion: Percutaneous radiofrequency ablation of locally advanced ade- when assessing IPMN at MRI-MRCP, show an interobserver agreement rang- nocarcinoma has a high technical success rate and is effective in cytoreduc- ing between substantial (0.663) and almost perfect (0.908) when applying the tion. Fukuoka criteria, confirming their usefulness for lesion evaluation standardiza - tion. SS 8.10 Application of Fukuoka radiological criteria in SS 8.8 pathologically confirmed IPMNs of the pancreas: is it Functional imaging of pancreatic adenocarcinoma using possible to differentiate between invasive/high- perfusion CT and diffusion-weighted imaging: moderate grade dysplasia IPMNs and noninvasive/ correlation with clinicopathological features low-grade dysplasia IPMNs? J. Kovac, A. Đuri ć-Stefanović, L. Lazić, T. Nikolić, S. Jevtić, C. Fabris, G.A. Zamboni, R. Negrelli, L. Bertuzzo, D.V. Vasin, D. Masulovic; Belgrade/RS G. Marchegiani, G. Mansueto; Verona/IT Purpose: To determine the value of perfusion CT and diffusion-weighted im- Purpose: IPMNs are intraductal cystic neoplasms with malignant potential. aging (DWI) in the evaluation of patients with pancreatic adenocarcinoma, and The Fukuoka criteria are commonly used to assess the risk of malignancy. Our to assess the correlation of perfusion CT parameters, and apparent diffusion purpose was to try to differentiate between high-moderate grade dysplasia coefficient (ADC) with clinicopathological features. IPMNs and low-grade dysplasia IPMNs using the mentioned criteria. Material and methods: Forty-four patients with histologically proven pancre- Material and methods: We reviewed the preoperative MRIs from 46 patients atic adenocarcinoma who underwent CT perfusion and DWI were prospec- (27 males, 19 females; mean age 63,7 years) with a confirmed pathologic diag - tively included in the study. Tumor blood flow (BF) and blood volume (BV) were nosis of IPMN. Two readers analyzed in consensus the MRIs for tumor size automatically calculated by commercial maximum slope based software. ADC (>/< 3 cm), diameter of the main pancreatic duct (MPD) (>/< 10 mm), presence was calculated for b 0, and 800 s/mm . BV, BF, and ADC values were com- of enhancing solid components, radiologic signs of pancreatitis, enhancing or pared using Mann-Whitney test between patients with pancreatic carcinoma non-enhancing nodules, thickened/enhancing walls, abrupt change in MPD and control group. Moreover, patients with pancreatic adenocarcinoma were caliber with distal atrophy, lymphadenopathy. The data were compared using classified into two groups according to the tumor grade (30 high-grade, and 14 Fisher’s test dividing the patients into two groups: high-moderate grade dys- low-grade lesions) and the difference of perfusion CT parameters and ADC plasia IPMNs (32) vs low-grade dysplasia IPMNs (14). was assessed. Correlation was tested using Spearman’s test. Results: MPD >10 mm (p=0.04), change in MPD caliber and distal atrophy -3 Results: The mean values of BV (ml/100g), BF (ml/min/100g), and ADC (x10 (p=0.037) and radiological signs of pancreatitis (p=0.008) were significantly mm /s) were significantly lower in pancreatic carcinoma (BV=4.41±1.84, more common in patients with high-moderate grade dysplasia IPMNs than in BF=24.03±12.81, ADC=1.028±0.19) than corresponding values of healthy patients with low-grade dysplasia. Enhancing solid components were present pancreatic tissue (BV=16.92±6.88, BF=68.71±11.13, ADC=2.312±0.532). A in 12/32 high-moderate-grade IPMNs and 1/14 low-grade IPMN (p=0.07). No statistically significant difference was found for BV, BF, and ADC between high significant differences were observed between the two groups for the other and low-grade tumors (all p˂0.05). Significant correlation was found among parameters evaluated. tumor diameter and BV (ρ=-0.445, p=0.029), M-stage and BV (ρ=-0.286, Conclusion: Fukuoka criteria appear to be helpful in differentiating between p=0.049), M-stage and ADC (ρ=0.274, p=0.038), ADC and BV (ρ=0.551, invasive/high-moderate grade IPMNs and low-grade IPMNs. The most useful p=0.012), ADC and BF (ρ=0.665, p=0.023). criteria are MPD >10 mm, change in MPD caliber and distal atrophy and radio- Conclusion: Perfusion CT and DWI provide an assessment of tumor vascular- logical signs of pancreatitis. ity and cellularity and can be used for preoperative prediction of high-grade pancreatic adenocarcinoma. S679 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 Liffey Hall 1 SS 9.2 Neurons vs neural networks: comparison of machine Scientific Session SS 9 learning/computational neural networks to radiologists Radiomics and artificial intelligence in in characterization of subcentimeter liver nodules in colorectal carcinoma found on staging CT scans abdominal imaging K. Khalili, R. Lawlor, M. Pourafkari, T.K. Kim, H.-J. Jang, M. Atri, S. Johnson, A. Martel; Toronto, ON/CA SS 9.1 Purpose: To determine if machine learning could improve assessment of “too Putting things in perspective: advanced image small to characterize” hepatic nodules in patients presenting with colorectal processing with Radiomics versus the radiologist’s “gut carcinoma (CRC). feeling” to predict treatment response in rectal cancer 1 1 1 Material and methods: Imaging of patients with CRC was assessed for the J.J.M. Van Griethuysen , M.J. Lahaye , M. Maas , 2 3 4 following inclusion criteria: local CT scan in portovenous phase, <1cm hepatic H.J.W.L. Aerts , F.C.H. Bakers , R.F.A. Vliegen , 1 1 1 nodule(s) & >2-year follow-up CT/MRI. 199 patients with 547 nodules were R.G.H. Beets-Tan , D.M.J. Lambregts ; Amsterdam/NL, 2 3 4 included. Malignancy was confirmed by ≥30% change, benignity by stability of Boston, MA/US, Maastricht/NL, Heerlen/NL >2 years. Machine learning (ML): predictive features from a pool of 100 ex- Purpose: To compare the predictive performance of advanced image pro- tracted texture features were used in a random forest classifier (200 trees, cessing (Radiomics) and visual morphologic evaluation by expert-radiologists 5-fold cross-validation) to develop diagnostic models from a training set of 150 to predict response to neoadjuvant chemoradiotherapy (nCRT) on pre-treat- patients (400 nodules). In addition, five convolutional neural networks (CNN) ment MRI of rectal cancer. were trained using a fivefold cross-validation technique. Results were aver - Material and methods: The pre-treatment MRIs (T2W + b1000/1100-DWI) of aged across folds for both ML & CNN. The remaining 147 nodules were used 130 patients were analyzed to predict the likelihood of a good (Mandard tu- as the test set to compare the performance of 3 abdominal radiologists to the mour regression grade (TRG) 1-2) or complete (TRG1/pCR) response, respec- better of ML versus CNN techniques, measured through receiver operating tively, according to two methods. First, two expert-radiologists estimated the characteristic (ROC) analysis. likelihood of a good or complete response, respectively, using a 5-point confi - Results: Prevalence of disease were 25.5% and 40.5% in the learning and dence score based on an overall visual morphologic assessment (size/shape/ testing sets. In the ROC analysis of training set, CNN (mean AUC 0.89, range border/signal, TN-stage, fascia invasion, extramural vascular invasion (EMVI)). 0.84-0.94) outperformed ML (mean AUC 0.78, range 0.69-0.83) and thus was Second, 3421 Radiomics-features were extracted using PyRadiomics. Data selected for comparison to radiologists. ROC analysis of the test set demon- were analyzed using bootstrap with 100 iterations, for each iteration, patients strated a significantly better performance for radiologists (mean AUC 0.94, were randomly split 70:30 (training:testing). In the training set, top 25 features range 0.91-0.95) than CNN (0.85, p=0.008). Using a threshold of 0.5, CNN were selected using minimum-Redundancy Maximum-Relevance (mRMR). obtained a sensitivity/specificity of 80.0%/78.4%. Selected features and the radiologists‘ scores were then fitted into separate Conclusion: CNN shows promise in the characterization of subcentimeter he- logistic regression models, which were evaluated on the testing set. patic nodules in CRC though not matching expert radiologists. Future work Results: 63 patients had a good response and 29 a complete response. To with larger training set is needed. predict a good response, average area under the ROC-curve (AUC) for the two expert-radiologists was 0.69 and 0.63, versus a maximum AUC 0.66 for the best-performing Radiomics-feature (ADC LoG-5mm-GLCM-IDN; a measure of heterogeneity). To predict a complete response AUC was 0.79 and 0.69 for the expert-radiologists versus AUC 0.72 for the best Radiomics-feature (ADC LoG-5mm-GLCM-IDN). Conclusion: Visual morphologic assessment of pre-treatment MRI by expert- radiologists results in comparable performance as (individual) Radiomics-fea- tures to predict response to neoadjuvant treatment in rectal cancer. S680 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 9.3 SS 9.4 Decreased expression of OATP8 in hepatocellular Radiomics as a novel tool for primary nodal staging in carcinoma evaluated by whole-lesion radiomics on rectal cancer 1 1 1 Gd-EOB-DTPA enhanced MRI J.J.M. Van Griethuysen , D.M.J. Lambregts , S. Trebeschi , 1 1 1 2 L. Cao, J. Chen, H. Jiang, B. Song; Chengdu/CN M. Maas , M.J. Lahaye , G.L. Beets , F.C.H. Bakers , 1 3 1 R.G.H. Beets-Tan , H.J.W.L. Aerts ; Amsterdam/NL, Purpose: To investigate the value of whole-tumor radiomic features on mul- 2 3 Maastricht/NL, Boston, MA/US tiphasic gadolinium-ethoxybenzyl-diethylenetriaminepentaacetic acid (Gd- EOB-DTPA) enhanced MRI in evaluating organic anion transporting polypep- Purpose: To assess the potential of quantitative MR-imaging analysis using tide 8 (OATP8) expression in HCC patients. Radiomics for nodal characterization in primary rectal cancer staging. Material and methods: 67 consecutive surgically confirmed HCC patients Material and methods: 226 lymph nodes (from 34 rectal cancer patients who were prospectively enrolled and underwent Gd-EOB-DTPA enhanced MR ex- underwent MRI followed by surgical resection) were analyzed and matched amination. Radiomic features on each sequence were derived using in-house node-by-node with histopathology. Short axis was measured for each node software (Analysis Kit, GE Healthcare). The expression intensities of OATP8 (on T2W-MRI) and nodes were manually segmented on T1W and T2W-MRI to were histopathologically determined and scored as: 0, no expression; 1, weak extract 748 radiomic features (using PyRadiomics). Using principal feature se- expression; 2, moderate expression; and 3, high expression. Lasso logistic lection, we selected the 25 features providing best complementary information regression was used for feature selection. Correlation analysis and nonpara- to routine nodal size-measurements. Performance to distinguish between ma- metric test were conducted accordingly. Receiver operating characteristic lignant and benign nodes was assessed with areas under the curve (AUC) (ROC) analysis was applied to determine diagnostic performances of selected per-feature using a binomial mixed effects logistic regression model, correct- features in evaluating the decreased OATP8 expression. ing for patient number. False detection rate (FDR) correction set at 10% was Results: The CorrelationAllDirection_offset1_SD, IDMAllDirection_offset4_SD applied to correct for multiple testing. and LongRunEmphasis_angle45_offset1 at T1-pre imaging, the IDMAllDirec- Results: 17/226 nodes were malignant. Selected features were compared to tion_offset1_S at portal venous phase (PVP), and the IDMAllDirection_offset1_ (and combined with) nodal size-measurements using a bootstrap method with SD and IDMAllDirection_offset4_SD at hepatobiliary phase (HBP) of HCC 100 iterations; the model was trained on a random subset of patients and showed significant correlation with OATP8 expression (p<0.05). All features tested on the remaining patients (train:test ratio 80:20). Average performance differed significantly between decreased expression (score 0-2) and high ex - for nodal size-measurements was AUC 0.82. The best performing Radiomic- pression (score 3) group (P<0.05 for all). The largest area under ROC curve feature after FDR correction (gray level co-occurrence matrix (GLCM)-differ- (AUC) of T1-pre images (IDMAllDirection_offset4_SD), PVP (IDMAllDirection_ ence average; a measure of heterogeneity) resulted in an AUC of 0.83; the offset1_SD), and T1-HBP (IDMomentAllDirection_offset4_SD) were combination of this feature with nodal size resulted in an AUC of 0.85. 0.69(CI[0.55, 0.81]), 0.66(CI[0.52, 0.78]) and 0.77(CI[0.63, 0.87]), respectively, Conclusion: Radiomics may provide valuable quantitative information to char- in identifying decreased OATP8 expression in HCC. T1-pre images (IDMAll- acterize rectal cancer lymph nodes; particularly features reflecting nodal het - Direction_offset4_SD) showed highest sensitivity (87.0%, CI[66.4%, 97.2%] vs erogeneity are promising. Although Radiomics slightly improves staging per- 78.3% CI[56.3%, 92.5%], 65.2% CI[42.7%, 83.6%]) while T1-HBP (IDMomen- formance, the added benefit compared to routine size-based staging appears tAllDirection_offset4_SD) showed highest specificity (80.7% ,CI[62.5%, to be limited. 92.5%] vs 48.4%, CI[30.2, 66.9], 51.6%, CI[33.1%, 69.8%]). Conclusion: Whole-tumor Radiomics features based on Gd-EOB-DTPA en- SS 9.5 hanced MRI, especially the IDM on HBP images, offers a potential avenue to- Diffusion kurtosis MRI for evaluating inflammatory ward preoperative evaluation of the decreased expression of OATP8 in HCC. activity in ulcerative colitis: pilot study J. Podgorska, K. Pasicz, E. Zagórowicz, B. Gołębiewski, P. Kuś, J. Jasieniak, P. Wieszczy, A. Anysz-Grodzicka, J. Pałucki, E. Fabiszewska, W. Skrzyński, P. Kukołowicz, A. Cieszanowski; Warsaw/PL Purpose: To assess the efficacy of diffusion kurtosis imaging (DKI) in the as - sessment of inflammatory activity of ulcerative colitis (UC). Material and methods: 17 patients with UC underwent 3T magnetic reso- nance DKI (b values of 0-2000 s/mm ) within a short time (1-6 days) of endo- scopic evaluation and with no bowel preparation. The inflammatory activity of up to 5 bowel segments (rectum, sigmoid, descending, transverse, ascending colon) was graded on endoscopy with Mayo score. Apparent diffusion for non- Gaussian distribution (DK ) and apparent kurtosis coefficient (K) on DKI as well as apparent diffusion coefficient (ADC) on diffusion weighted imaging (DWI) were correlated with Mayo score. T-test was used to compare the groups. Results: In total 54 bowel segments: endoscopically inactive (n=18) and with mild to severe activity (Mayo 1-3) (n=36) were analysed. Statistically significant differences between inactive and active disease were found in DK (mean = -3 2 -3 2 2.3x10 mm /s, standard deviation (SD) = 0.61x10 mm /s and mean = -3 2 -3 2 1.81x10 mm /s, SD = 0.51 x10 mm /s respectively, p=0.003) and ADC (mean -3 2 -3 2 -3 2 = 1.3 x10 mm /s, SD = 0.27x10 mm /s and mean = 1.15x10 mm /s, SD = -3 2 0.23x10 mm /s respectively, p=0.041). No significant difference was found in K. Overlapping of parameters was less significant in DK than in ADC. Conclusion: DKI of UC correlates with disease activity might be superior to Gaussian DWI model for assessment of the inflammatory activity. SS 9.6 withdrawn by the authors S681 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 9.7 SS 9.9 Consistency of MR radiomic features on T2-weighted Multiparametric combined fluorodeoxyglucose-positron imaging of the liver emission tomography (FDG-PET)/CT and MR imaging to J. Santinha, J.M.G. Lourenço, I. Santiago, C. Matos, predict response to chemoradiotherapy in rectal cancer: N. Papanikolaou; Lisbon/PT whole tumor versus sub-volume analysis 1 2 1 N. Schurink , M. Berbee , J. Van Griethuysen , Purpose: To assess the intra-observer, inter-observer and test-retest repeat- 2 1 1 2 W.J.C. Van Elmpt , M. Maas , M.J. Lahaye , F.C.H. Bakers , ability of radiomic features on T2-weighted magnetic resonance imaging of the 1 1 1 R.G.H. Beets-Tan , D.M.J. Lambregts ; Amsterdam/NL, liver. Maastricht/NL Material and methods: Two axial 2D turbo-spin-echo T2-weighted abdominal acquisitions were acquired, with a minimum interval of 15 minutes between Purpose: To assess the individual and complementary performance of quanti- acquisitions, in 14 different patients (8 men; mean age=59.5 years) performing tative parameters from pre-treatment multiparametric MRI and FDG-PET/CT an abdominal MRI examination, in a „scan-rescan“ design. Examinations were to predict treatment response in rectal cancer. performed in a 1.5T Ingenia Philips scanner with the following parameters: Material and methods: A pilot group (n=20) underwent multiparametric MRI slice thickness=5mm; gap=1mm; echo time=80ms; repetition time=850ms; flip (T2W + diffusion weighted imaging (DWI); b0,100,500,1000) and FDG-PET/CT angle=90º; Bandwidth=507Hz; echo train lenght=72. Two radiologists manu- before neoadjuvant treatment + surgery. Images were anatomically co-regis- ally segmented the whole liver (VOI) twice in both acquisitions from each pa- tered using rigid + non-rigid registration. Whole-tumor volumes were segment- tient. A total of 2247 radiomic features (shape, first order, glcm, glrlm, glszm, ed on T2W-MRI and transferred to the other modalities/sequences; each tu- gldm and ngtdm of the original and filtered images - exponential, logarithm, mor was additionally divided into 7mm sub-volumes. The following parameters square, square root, two levels of wavelet and Laplacian of Gaussian with 1-,3- were calculated per-tumor and sub-volume: T2-texture (uniformity/entropy), , 5-mm) were extracted from each VOI using the Pyradiomics package. The T2-signal intensity (SI), apparent diffusion coefficient (ADC) (mean/max/min/ th th intra-observer, inter-observer, and test-retest repeatability were assessed us- median/10 & 90 percentile), standard uptake value (SUV) (mean/max/medi- ing the intraclass correlation coefficient (ICC). An ICC≥0.81 was considered an), CT-Hounsfield units (HU). Performance to predict poor response (=Mand - excellent. ard tumor regression grade (TRG)3-5) vs. good response (TRG1-2) was calcu- Results: 1084 features for Reader 1 and 1053 features for Reader 2 showed lated using receiver operating characteristic (ROC)-analysis for [1] each excellent intra-observer repeatability (mean=47.5%) with 1017 (mean=95.2%, individual parameter (whole-tumor), [2] multiparametric combination of the 5 45.3% of total) overlapping features. 1090 (48.5%) features showed excellent best-performing parameters (whole-tumor), and [3] the proportion (%) of inter-observer repeatability and 1258 (56.0%) features showed excellent test- ‘poor-response’ sub-volumes within the tumor. Sub-volumes were defined as retest repeatability. A total of 930 (41.4%) features (wavelet filters and origi - ‘poor response’ when ≥4 (out of 5) parameters within that sub-volume were nal≥100, firstorder=192 and glcm=241) showed excellent repeatability for in - indicative of TRG3-5, using a cut-off derived from dichotomisation by median tra-observer, inter-observer and test-retest repeatability. split. Conclusion: Our results indicate that, for T2-based whole-liver analysis and Results: Best single predictive parameters were T2-uniformity (area under after accounting for intra-observer, inter-observer and test-retest variability, curve (AUC)0.78), T2-SI (AUC0.70), ADCmax (AUC0.76), SUVmax (AUC0.78) less than half of radiomics features preserve an excellent repeatability. These and SUVmean (AUC0.73). Combined multiparametric performance (whole-tu- „stable“ features may, therefore, be more „trustworthy“ for T2-based whole- mor) was AUC0.91. The proportion of ‘poor-response’ sub-volumes resulted in liver radiomics analysis. AUC0.74. Conclusion: Multiparametric analysis of quantitative MRI and FDG-PET/CT data has potential added value to predict response to neoadjuvant treatment. SS 9.8 If there is a potential benefit for performing sub-volume (or voxel-wise) analy - Texture analysis of preoperative CT images for sis, this needs to be established by further and larger studies; with our current prediction of liver volume regeneration after major method, we have so far not demonstrated a clear added value. hepatectomy in living donor transplantation 1 1 1 2 1 J.E. Kim , J.H. Kim , S.J.H. Park , S.-Y. Choi , J.S. Bae , 1 1 1 2 S.J. Jeon , J.K. Han ; Seoul/KR, Bucheon/KR Purpose: To predict the rate of liver regeneration after donor hepatectomy us- ing preoperative laboratory test and computed tomography (CT) texture analy- sis in living donor transplantation. Material and methods: 112 living donors (M:F = 79:33, 31 years) who per- formed preoperative CT were included in this study. After transplantation, fol- low-up CT was performed after median 123 days. We measured the volume of future remnant liver (FLR) on preoperative CT and volume of the remnant liver (LR) on follow-up CT. Regeneration index (RI) was calculated using following equation: [(V -V )/V ] × 100. Texture analysis regarding FLR was semiauto- LR FLR FLR matically performed. We used a stepwise multivariable regression to assess associations of laboratory results and texture parameters in relation to RI and make the best fit predictive model. Results: The mean RI was 110.7 ± 37.8%, highly variable ranging from 22.4% to 247.0%. Among texture parameters, standard deviation, variance, volume of FLR and grey level co-occurrence matrices (GLCM) contrast were found to have significant correlations between RI in univariable analyses. In a multivari - able analysis adjusting for other factors, a smaller volume of FLR (ß -0.17, 95% CI -0.22 to -0.13) and lower GLCM contrast (ß -1.87, 95% CI -3.64 to -0.10) were associated with higher RI. The regression equation predicting RI was fol- lowing: RI = 203.82 + 10.42 x preoperative serum total bilirubin (mg/dL) – 0.17 x V (cm ) – 1.87 x GLCM contrast (×100). FLR Conclusion: In our study, the volume of FLR and GLCM contrast were inde- pendently associated with the rate of liver regeneration. CT texture analysis can be useful in predicting liver regeneration for the donors undergoing right hepatectomy. S682 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 11:00 - 12:30 Wicklow Hall 2A SS 9.10 Academic-industry collaborations involving quantitative Scientific Session SS 10 imaging biomarkers in drug development clinical trials M.S. Middleton, J. Cui, W.C. Henderson, J.C. Hooker, Liver: focal and diffuse liver pathologies, G. Hamilton, N. Szeverenyi, C. Sirlin; La Jolla, CA/US intervention and transplantation Purpose: To describe an academic-industry collaboration paradigm devel- oped over ten years while acting as a Radiology Coordinating Center (RCC) for SS 10.1 drug development clinical trials involving quantitative imaging biomarkers Gadoxetic acid-enhanced MRI before primary liver (QIBs) such as MRI hepatic proton density fat fraction (PDFF), magnetic reso- transplantation to predict recurrence of HCC nance elastography (MRE) liver stiffness, and magnetic resonance spectros- S. Lee, K.W. Kim, W.K. Jeong; Seoul/KR copy (MRS) fat tissue spectral characteristics. Our role as RRC for these com- Purpose: We performed a multicenter retrospective survey to investigate the plex, collaborative studies required iterative refinement of project planning, prognostic value of preoperative gadoxetic acid-enhanced MRI in predicting workflow, and management. HCC recurrence after liver transplantation (LT). Material and methods: We describe non-disclosure agreements and con- Material and methods: From two large LT centers, data was collected from tracts, statements of work and budgets, regulatory issues, conflicts of interest, 140 recipients with HCC who underwent primary LT and pretransplant gadox- site selection and training, intake quality control (QC), analysis, analysis QC, etic acid-enhanced MRI between January 2009 and December 2013. results reporting, protocol deviations and violations, and standard operating Results: The 1-, 3-, and 5-year recurrence-free survival rates were 91.3%, procedures. 85.4%, and 82.4%, respectively. Multivariate analysis revealed that being out- Results: We have acted as an RCC for 21 industry clinical trials involving liver side the Milan criteria (HR, 3.062; 95% CI, 1.309–7.159; P = 0.010) and peritu- QIBs, and four industry clinical trials involving MRS of the abdomen, thigh, and moral hypointensity on hepatobiliary phase (HBP) (HR, 10.268; 95% CI, 4.248– lower leg muscles. These studies were conducted at over 400 sites worldwide 24.820; P < 0.001) were independent predictors of HCC recurrence. The involving over 5,000 MR exams using the five main MR manufacturers for accuracy in categorization of the Milan criteria on pretransplant MRI was PDFF and the three MR manufacturers with MRE capability. Collectively, these 92.9% in correlation with the explanted liver. Peritumoral hypointensity on HBP exams have involved almost every conceivable combination of scanner type was significantly associated with worse tumor grade and microvascular inva - and sequence parameter settings appropriate for the measured QIBs. We have sion (P = 0.038 and P < 0.001, respectively). Using a combination of the Milan developed robust templates for an imaging manual, an RCC charter, and billing criteria and peritumoral hypointensity on HBP, preoperative MRI further strati- log and results-reporting spreadsheets. fied the risk of tumor recurrence after LT (P < 0.001), and 5-year recurrence- Conclusion: The methods, templates, and other study materials developed in free survival rates for patients outside of the Milan criteria and non-peritumoral collaboration with the pharmaceutical industry have helped us plan and man- hypointensity on HBP were comparable to those within the Milan criteria age new industry studies involving QIBs. Many of these concepts may be gen- (89.0% vs. 75.9%) (P = 0.090). eralizable to other core labs and to other clinical trials using quantitative bio- Conclusion: Pre-transplant MRI using gadoxetic acid offers not only precise markers, within and outside the field of radiology. morphologic tumor stage, but also additional biologic tumor behavior on the risk of HCC recurrence after LT. Patients outside of the Milan criteria with non- peritumoral hypointensity on HBP may achieve acceptable outcomes compa- rable to that of patients within Milan criteria. SS 10.2 Radiological-pathological correlation of pre-transplant imaging and explanted livers: American College of Radiology (ACR) Liver Imaging Reporting and Data System (LI-RADS) 2017 category versus Organ Procurement and Transplantation Network (OPTN) class in the diagnosis of HCC C.M. O Brien, A.C. O‘Brien, P.A. McCormick, N. Nolan, S.J. Skehan, R. Ryan, D.E. Malone, R. Gibney; Dublin/IE Purpose: To evaluate concordance of the LI-RADS 2017 and OPTN-United Network for Organ Sharing (UNOS) systems in the diagnosis of hepatocellular carcinoma (HCC). Material and methods: Liver transplant (OLT) patients with a pre-operative diagnosis of HCC and explants with HCC on pathology from 01/01/2014 to 09/09/2017 were retrospectively identified using the OLT database. Pre-OLT CT/MRI reports were reviewed. In patients with LI-RADS 5, 4/5, 4 or 3 lesions, the descriptions in the reports were used to categorize the lesions using the LI-RADS 2017 (LR) and OPTN-UNOS systems and the histopathology coding of explanted livers was reviewed. Data were collated in an MSExcel spread- sheet and analysed on a per patient basis. Results: 49 OLT patients met selection criteria. LR and OPTN scores were assigned to all patients. 39/49 with HCC were LR-5. Of these, 38/49 were OPTN-5, 1 did not meet arterial enhancement criteria and in retrospect was an LR-4 lesion. OPTN subclasses were: 5B 16; 5A 9; 5A-g 4; 5X 2; 5T 7. LR-4 or LR-4/5 nodules were present in 5/49 and 1/49 patients with proven HCC. 3/6 had prior TACE and were OPTN-5T. 3/6 were not OPTN 5, all patients had HCC. 4/49 were LR-3; none were OPTN 5. Conclusion: No discrepancies were found between the 2017 ACR LI-RADs 5 and OPTN-UNOS 5 lesion classification systems in these HCC patients. Me - ticulous adherence to technical and interpretative aspects of both systems is essential. An advantage of LI-RADS vs OPTN-UNOS is that LI-RADS 4 (prob- able HCC) lesions not meeting OPTN 5 criteria are identified for close follow- up. S683 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 10.3 SS 10.5 The relationship between the location of the portal vein Contrast-enhanced oncologic staging MRI of abdomen thrombus, the patency of the portal veins, and the and pelvis with a hepatocyte-specific agent in a large- coexisting disease bore MR scanner: optimizing extrahepatic tumor K. Ibukuro, G. Ogasawara, H. Fukuda, K. Tobe, M. Kishino; evaluation in neuroendocrine neoplasms 1 1 1 2 Tokyo/JP U. Fehrenbach , U. Fahlenkamp , V. Prasad , M. Pavel , 1 1 1 2 D. Geisel , T. Denecke ; Berlin/DE, Erlangen/DE Purpose: To clarify the relationship between the location of the portal vein thrombus, the patency of the portal vein, and the coexisting disease. Purpose: Liver-specific MR contrast agents in neuroendocrine neoplasms Material and methods: We retrospectively reviewed 48 patients with portal (NEN) have shown its advantages in evaluation of hepatic tumor burden. How- vein thrombus observed on contrast-enhanced CT scans for six years. The ever, a standard scan protocol is insufficient in extrahepatic and especially coexisting diseases were classified as (A) liver cirrhosis (LC) (n=29) and (B) in - pelvic tumor evaluation. The aim of the study is to evaluate an MRI shuttle flammatory diseases (Inf) including cholangitis or cholecystitis (n=13) and he - protocol with fast 3D-T1w sequences and hepatocyte-specific contrast (Gd- patic abscess (n=6). The locations of the thrombus and the patency of the EOB) to enable imaging abdomen and pelvis in optimal first-pass contrast portal vein were classified as peripheral (second-order division of intrahepatic phases in the setting of a large-bore MRI scanner with shorter axial field of portal vein, n=18) and central (n=30), complete obstruction (n=20) and patent view. (n=28), respectively. We reviewed the presence of ascites, esophageal varices, Material and methods: 96 patients with NEN of the abdomen were scanned gastro- or spleno-renal shunt, and Child classification. The logistic regression with the first pass (FP) pelvic protocol in a Gd-EOB MRI. FP pelvic sequence analysis was used to identify the statistically significant characteristics for the was acquired in between portal venous and venous liver phase. We compared location of the thrombus and the patency of the vein. The level of significance pelvic vessel and tissue enhancement between FP and a standard equilibrium was set to 5%. The Cramér’s V was measured for the association between the (EQ) pelvic sequence. Dynamic liver enhancement with the new protocol in- location, the patency, and the coexisting disease. cluding FP pelvic sequence was compared to a standard protocol without ad- Results: The coexisting disease was the primary factor that affected the loca- ditional sequence. tion of the thrombus (odds ratio: 32; LC vs. Inf) and the patency (odds ratio: 18; Results: The FP pelvic sequence showed improved vessel and tissue en- LC vs. Inf). The V were as follows: 0.693 (the coexisting disease and the loca- hancement. Aortic (140.81 vs 116.43) and venous (89.92 vs 62.76) vessel en- tion), 0.612 (the coexisting disease and the patency), 0.306 (the location and hancement in the lower abdomen showed significantly (p=0.000) higher signal the patency). intensities in FP pelvic than in the EQ phase. The FP sequence also showed Conclusion: The coexisting disease had the most important role for the loca- significantly improved lymph node enhancement (57.84 vs 41.15; p=0.000). tion of the portal vein thrombus and the patency of the portal vein. Dynamic liver enhancement in the FP protocol showed no significant differ - ences to the standard protocol (57.48 vs 53.01; p=0.143). Conclusion: Including a FP pelvic sequence in between portal venous and SS 10.4 venous liver phase by shuttling of the patients’ table enables improved tumor Assessment of liver function in patients with HCC using evaluation of the pelvis region in Gd-EOB MRI. Additionally, there was no qual- perfusion-weighted MRI with gadolinium-ethoxybenzyl- ity loss in liver enhancement in the proposed FP pelvic protocol. diethylenetriamine pentaacetic acid T. Duan, J. Chen, X. Lin, B. Song; Chengdu/CN SS 10.6 Purpose: To prospectively evaluate liver function in patients with HCC using MRI assessment of transarterial chemoembolization perfusion-weighted magnetic resonance imaging (PW-MRI) with gadolinium- effects in patients with neuroendocrine hepatic ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA). metastasis Material and methods: PW-MRI was performed on 41 HCC patients on a 3.0 M. Lapteva, M.A. Shorikov, D. Frantsev, O. Sergeeva, Tesla MR scanner. The perfusion data were acquired using a prototype radial E. Virshke, B. Dolgushin; Moscow/RU stack-of-stars 3D spoiled gradient echo pulse sequence with golden-angle radial sampling schemes over the course of 6.25 minutes. Post-processing of Purpose: The purpose of the paper is to develop suitable transarterial chem- PW-MRI data was performed on an in-house software O.K. (Omini-Kinetics). A oembolization effects (TACE) evaluation criteria in liver metastatic neuroendo- 2-compartment 2-input pharmacokinetic model was chosen to generate crine tumor (mNET) patients on MRI. haemodynamic curving. We applied an extended Toft‘s linear model to gener- Material and methods: 23 liver mNET patients (G1-9, G2-10, G3-2, unknown ate the maps of K , K , V and the semi-quantitative parameters of time-to- grade 2) who underwent 36 TACE procedures were enrolled in the study. 68 trans ep e peak (TTP), max concentration and max slope. Perfusion parameters and vol- MRI studies performed 3-86 days before (median 28 days) and 23-147 days umes of normal hepatic parenchyma were conducted, then the products of (median 61 days) after TACE were analyzed retrospectively. Using 1.5T MRI- perfusion parameters and volume were calculated. Indocyanine green reten- scanner, we acquired: 1) 3DT1WIFS in axial plane before i.v. contrast injection tion at 15 min (ICG R-15) and model for end-stage liver disease (MELD) scores and after in portal and delayed phases; 2) diffusion-weighted images with ap- were also recorded. PW-MRI parameters, volumes and products were corre- parent diffusion coefficient (ADC) maps. We measured before/after TACE: 1) lated with both ICG R-15 and MELD scores using Spearman rank correlation maximum necrotic/fibrotic zone diameter and solid component thickness on analysis. its periphery in the largest targeted lesion; 2) ADC and MR contrast agent ac- Results: The product of TTP and volume showed poor correlations with ICG cumulation dynamics (dSl) in regions of interest in tumor solid component. R-15 (rho=-0.367, p=0.046). Meanwhile, K (rho=0.496, p<0.001), K Results: TACE effect according to response evaluation criteria in solid tumors trans ep (rho=0.930, p<0.001), max slope (rho=0.342, p=0.033), product of K and (RECIST)(1.1) was evaluated as progressive disease in 8.3% cases, stabiliza- trans volume (rho=0.898, p<0.001), product of K and volume (rho=0.898, p<0.001) tion in 83.3% cases and partial response in 8.3% cases. The diameter of the ep and product of max slope and volume (rho=0.388, p=.015) correlated signifi - fibrosis/necrosis areas before/after TACE demonstrated no changes (p=0.614), cantly with MELD scores. but solid component thickness on its periphery decreased significantly Conclusion: Measures of liver function obtained by Gd-EOB-DTPA PW-MRI (p=0.003, median 16%). ADC values increased significantly in solid portions of with tracer kinetic modelling may provide a suitable method for the evaluation mNETs after TACE (p=0.003). MR contrast agent accumulation of the lesions in of liver functional reserve. portal phase decreased significantly after TACE (p=0.003, median 8.5%). De - crease of MR contrast agent accumulation in a solid part of the target lesions more than 10% after TACE is associated with a longer time to progression of the disease (p=0.025). There is an increase in MR contrast agent wash-out time in the solid part of the target lesions after TACE (p=0.032). Conclusion: Parameters changed in liver mNET after TACE: 1) solid compo- nent thickness in cases of central necrosis/fibrosis (decrease); 2) ADC (in - crease); 3) ndSI T1WI in portal phase (decrease) and this is the positive prog- nostic factor; 4) MR contrast agent wash-out time (increase). S684 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / THURSDAY, JUNE 14, 2018 SS 10.7 SS 10.9 Differentiation of focal nodular hyperplasia from HCC: Gadobenate dimeglumine-enhanced MRI in patients the role of triple-phase arterial MRI with ascitic effusion 1 2 1 2 M. Gatti, R. Faletti, S. Rousset, L. Bergamasco, P. Fonio; M. Bonatti , G.A. Zamboni , F. Lombardo , R. Valletta , 1 1 2 Turin/IT G. Bonatti ; Bolzano/IT, Verona/IT Purpose: To determine the value of triple-phase arterial imaging to differenti- Purpose: To evaluate the pharmacokinetics of gadobenate dimeglumine (Gd- ate focal nodular hyperplasia (FNH) from HCC. BOPTA) in patients with ascitic effusion and to correlate it with the cause of the Material and methods: The study population included 52 patients who had effusion. arterial enhancing hepatic lesions (a total of 72 lesions: 41 HCC and 31 FNH). Material and methods: We included in our institutional review board ap- All patients underwent triple-phase arterial phase MRI using hepatocyte-spe- proved retrospective study 32 consecutive patients with ascites who under- cific agents (Gd-EOB-DTPA). Images were reviewed by two radiologists in went Gd-BOPTA-enhanced liver MRI (including hepatobiliary phase) on a 1.5T consensus: contrast enhancement ratios (CER), liver-to-lesion contrast ratios scanner in our Institution between January 2012 and December 2017. Clinical (LLC) and signal intensity (SI) were measured. The lesions were categorized data were collected. Appearance of ascitic fluid (hypo-, iso- or hyper-intense based on the peak of LLC into the following groups: 1) angiographic, 2) early in comparison to paravertebral muscle) was assessed on native and hepatobil- arterial and 3) late arterial. Data were analysed with Wilcoxon signed-rank test. iary phase gradient-recalled echo (GRE) T1-weighted images. Signal intensity Results: There was no difference in CER between FNH and HCC patients of ascites and paravertebral muscle was measured by means of a 1cm round (p>0.09). The SI increases through the different arterial phases 1) 242.08 --> region of interest (ROI) on both phases; ascites/muscle signal-to-noise ratio HCC 2) 251.84 --> 3) 253.95; however, the SI started to decrease in the late arte- (SNR) was calculated. FNH rial phase 1) 303.21 --> 2) 326.03 --> 3) 323.72. LLC were significantly Results: The cause of ascites was portal hypertension in 23/32 patients, peri- FNH higher than LLC in the angiographic phase (41.3 vs. 25.0; p=0.01), whereas toneal carcinomatosis in 6/32 and recent major surgery in 3/32. On native im- HCC there were not different in the other phases (p=0.20 and p=0.82, respectively), ages, ascitic fluid was hypointense in 90.6% of the cases and isointense in this behaviour was paralleled by the subgroup divisions: 17 (55%) FNH were 9.4%, whereas in the hepatobiliary phase it was hypointense in 3.1% of the st nd rd classified as the 1 group, 11 (35%) in the 2 and only 3 (10%) in the 3 ; on cases, isointense in 21.9% and hyperintense in 75% (p<0.001). Median as- the other hand, the HCC was homogeneously distributed (34%, 32% and cites/muscle SNR was 0.54 in native phase and 1.59 in hepatobiliary phase 34%). with a median increase of 269% (range 158-748%) (p<0.001). Median SNR Conclusion: This study highlights the different enhancement patterns of HCC increase was 294% in patients with portal hypertension, 194% in patients with and FNH: FNH resulted more conspicuous in the angiographic phase and peritoneal carcinomatosis and 239% in those after surgery (p>0.05). started its “wash out” in the late arterial phase; therefore, a multi-phase arte- Conclusion: Gd-BOPTA is excreted in ascitic fluid, which significantly increas - rial phase MRI may be an adjunctive tool to further differentiate HCC from es its signal intensity in hepatobiliary phase. No statistically significant differ - FNH. ences were found between the different types of ascites. SS 10.8 SS 10.10 Reappraisal of rate and risk factors of complications Evaluation of texture analysis for the differential after US-guided percutaneous liver biopsy: a diagnosis of focal nodular hyperplasia from retrospective analysis of 2405 biopsies hepatocellular adenoma on contrast-enhanced CT A. Maheux, S. Harguem, V. Vilgrain, M. Ronot; Clichy/FR images 1 2 2 2 R. Cannella , A. Borhani , B. Rangaswamy , M.I. Minervini , Purpose: To reappraise the rate and risk factors of complications after US- 2 2 1 2 A. Tsung , A. Furlan ; Palermo/IT, Pittsburgh, PA/US guided liver biopsy in a large single-center series of 2405 biopsies. Material and methods: We analyzed 2405 liver biopsies performed in 2137 Purpose: To investigate the value of texture analysis for the differentiation of patients (58% males, mean age 54±15 yo) between January 2008 and Decem- focal nodular hyperplasia (FNH) from hepatocellular adenoma (HCA) on con- ber 2013. Biopsies were performed for the characterization of focal liver le- trast-enhanced CT imaging. sions or to assess suspected or known chronic liver disease. Clinical, labora- Material and methods: This is a retrospective, institutional review board-ap- tory, and technical data were recorded for all biopsies. For focal liver lesions, proved study conducted in a single institution. A search of the medical records the following elements were also recorded: largest diameter, location, en- between 2008 and 2017 revealed 48 patients (47 females, 1 male) with 70 HCA hancement characteristics and pathological results. Occurrence of post-biop- and 50 patients (48 females, 2 males) with 62 FNH. All lesions were histologi- sy significant symptoms (any complaint motivating complementary imaging) cally proven and with available pre-operative CT imaging. Late hepatic arterial and complications were noted. Complications requesting specific treatment phase (HAP) and portal venous phase (PVP) images were used for texture (embolization or surgery) were considered as severe. analysis. A region of interest was placed on the largest tumor cross section Results: 1283 (53%) biopsies targeted focal lesions and 1122 (47%) were per- and in the non-lesional right hepatic lobe. Textural features were extracted us- formed for liver disease. Significant symptoms occurred after 134 biopsies ing a commercially available research software (TexRAD) which applies a (5.6%); the most frequent being pain (109/134). Complications occurred after 2-step filtration-histogram approach. The differences between textural param - 38 biopsies (1.6%); the most frequent being hemoperitoneum (n=21), subcap- eters of FNH and HCA were assessed using the Mann-Whitney U test. A p sular liver hematoma (n=11) and liver hematoma (n=6). One patient died. Se- value <0.05 was considered statistically significant. vere complications were present in 13 patients. In univariate analysis, pro- Results: On HAP images, mean, standard deviation (SD), entropy, mean of thrombin time (p=0.006), creatinine serum level (p<0.001), lesion largest positive pixels (MPP) and skewness were significantly higher in FNH than in diameter (p<0.001) and tumor pathology (p=0.040) were associated with the HCA on both unfiltered (p≤0.006) and filtered analysis (p≤0.02). On PVP mean, occurrence of complications but not platelet count. In multivariate analysis, MPP, skewness and kurtosis in FNH were significantly higher than in HCA only the lesion largest diameter was retained (OR 1.014 [1.002-1.026], (p≤0.006) on unfiltered images, while mean, SD and entropy were significantly p=0.018). higher in FNH on filtered images (p≤0.03). The analysis of the non-lesional liver Conclusion: Rate of complications after US-guided liver biopsy was low. Le- did not show any significant difference between two groups. sion largest diameter appears as the main risk factor of complication. Conclusion: Textural analysis parameters are significantly different between FNH and HCA. Imaging-based quantification of lesion heterogeneity may help in non-invasive differentiation of FNH from HCA. S685 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 The Auditorium SS 11.2 Differentiation of pancreatic neuroendocrine tumor Scientific Session SS 11 grade comparing 3D CT texture analysis and relative CT Pancreas: focal and diffuse pancreatic enhancement ratios 1 1 1 2 A. Sarno , G. Tedesco , V. Ciaravino , N. Cardobi , pathologies including endocrine tumours 2 2 1 R. De Robertis , P. Tinazzi Martini , L. Landoni , 1 1 1 1 S. Cingarlini , A. Scarpa , M. D‘Onofrio ; Verona/IT, SS 11.1 2 Peschiera del Garda/IT Additional value of tumour growth rate in patients Purpose: To evaluate the added value of 3D CT texture analysis compared to diagnosed with well-differentiated neuroendocrine CT enhancement ratios in the study of pancreatic neuroendocrine tumors tumours 1 2 3 4 (pNETs). C. Dromain , A. Sundin , P.S. Najran , H. Vidal Trueba , 5 6 7 8 Material and methods: 100 patients with pNETs were included in this study. L. De Mestier , J. Crona , M. Opalinska , L. Carvalho , 8 3 1 1 4 All tumors were pathologically diagnosed after resection or by means of bi- R. Franca , P. Borg , N. Vietti Violi , N. Schaefer , C. Lopez , 8 3 8 9 5 opsy, so histological grade was available in all cases. There were 31 G1, 52 G2, D. Pezzutti , A. Lamarca , F. Costa , M. Pavel , M. Ronot ; 1 2 3 and 17 G3 neoplasms with the presence of metastases in 53/100 (53%) cases. Lausanne/CH, Stockholm/SE, Manchester/UK, 4 5 6 7 3D CT texture analysis and a quantitative analysis of enhancement of CT stud- Santander/ES, Clichy/FR, Uppsala/SE, Krakow/PL, 8 9 ies were performed comparing the results with the tumor grading. CT texture Sao Paulo/BR, Erlangen/DE analysis results (mean value, variance, skewness, kurtosis) and relative CT en- Purpose: To assess the impact of lesion measurements and inter-reader re- hancement ratios were compared using Wilcoxon–Mann–Whitney correlation producibility on tumour growth rate (TGR) (% change in tumor volume/month) test. to predict patient outcome. Results: Included pancreatic neoplasms were located in the head of pancreas Material and methods: Baseline and 3-month (+/-1) images from patients (42%), in the body–tail (52%) and 6 patients had entire involvement of the with advanced, grade 1-2 neuroendocrine tumours (NETs) were retrospective- pancreatic gland (6%). CT texture analysis and CT quantitative analysis were ly reviewed by 2 readers for TGR calculation. Influence of number, location of 3m done in all cases. Among the three grading tumors groups (G1, G2 and G3) lesions and tumor burden on TGR on patient outcome (measured as pro- 3m kurtosis parameter resulted in statistically significant difference (p < 0.05) at CT gression-free survival (PFS)) was assessed by uni/multivariable Cox regression texture analysis. Instead, there was no statistically significant difference in CT analysis. Spearman correlation and Kruskal-Wallis test were employed. Agree- enhancement ratios between G1 and G2 groups at CT quantitative analysis. ment between readers was assessed by the Lin’s concordance coefficient Conclusion: CT texture analysis parameters can be more suitable to differenti- (LCC) and kappa (KC). ate the three grading pNETs, compared to relative CT enhancement ratios. Results: A total of 790 lesions measured in 217 patients were included. Me- dian PFS was 22.9 months. On univariable analysis, number of lesions (</=>4), SS 11.3 tumor burden and presence of liver metastases were significantly correlated to Location of liver metastases based on the site of PFS. On multivariate analysis, ≥4 lesions (HR:1.89 (95%CI:1.01-3.57)), TGR 3m primary pancreatic neuroendocrine tumour ≥0.8%/month (HR:4.01 (95%CI:2.31-6.97)) and watch-and-wait treatment cor- C. Fabris, A. Bissoli, M.C. Ambrosetti, G.A. Zamboni, related with shorter PFS. No correlation was found between TGR and num- 3m G. Mansueto; Verona/IT ber of lesions employed for TGR calculation (rho: -0.2; p value: 0.1930). No difference in mean TGR across different organs was shown (p value: 0.6). Purpose: The presence of liver metastases from neuroendocrine tumors does 3m Concordance between readers was acceptable (LCC: 0.52 (95%CI: 0.38- not preclude surgery. Knowing the distribution of these metastases is impor- 0.65); KC: 0.57; agreement: 81.55%). TGR3m remained a significant prognos - tant for treatment planning. Our purpose is to investigate whether the distribu- tic factor when data from the second reader were employed (HR: 4.35 (95%CI: tion of liver metastases from pancreatic neuroendocrine tumors (pNETs) is in- 2.44-7.79); p value<0.001) and regardless of second reader expertise (HR: fluenced by the “streamline phenomenon” within the portal vein. 1.21 (95%CI: 0.70-2.09); p value: 0.493). Material and methods: Institutional review board approval was waived for this Conclusion: Our results suggest that number of measured lesions and TGR retrospective study. We reviewed retrospectively the CTs performed on 49 pa- 3m value impact on PFS of patients with advanced NETs. The number of target tients with pNETs (27 males, 22 females) and liver metastases. Patients were lesions does not seem to affect the performance of TGR calculation. In addi- divided into 2 groups according to the pNET site: group A (11 patients) with a tion, TGR3m role as a prognostic factor was maintained regardless of reader’s pNET in the head (mean age 50.3 years); group B (38 patients) with a pNET in expertise. the body–tail (mean age 59,3 years). Two readers in consensus evaluated tu- mor site, diameter, vascular invasion and number of metastases within each lobe. Student’s and Fisher’s tests were used as appropriate. Results: Primary tumors in the body–tail (group B) were larger than those in the head (group A) (59.55 vs 39.82 mm; p=0.043). The splenic vein was more commonly involved in patients in group B. The number of liver metastases was higher in the right lobe in both groups (p=ns). The ratio of metastases in the right-to-left hemi-liver was 2.5:1 for group A compared with 1.79:1 for group B (p=0.0423). Conclusion: As expected, tumors from all locations in the pancreas will give a higher number of metastases in the right lobe. The right-to-left ratio is, how- ever, significantly different based on the site of the primary tumor. S686 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 11.4 SS 11.6 Pancreatic neuroendocrine tumor grade discrimination Pancreatic steatosis is associated with restriction of with enhancement pattern in the arterial phase: a pancreatic exocrine function 1 2 1 1 meta-analysis J.-P. Kühn , M.-L. Kromrey , R.-T. Hoffmann , M. Laniado , 2 1 2 H. Yu, M. Li; Chengdu/CN N. Friedrich ; Dresden/DE, Greifswald/DE Purpose: To assess the diagnostic test accuracy for the discriminative poten- Purpose: Recent research is now focused on the clinical significance of pan - tial of hypo-enhancement in the arterial phase to differentiate high-grade pan- creatic steatosis. The purpose was to investigate if pancreatic steatosis is as- creatic neuroendocrine tumors (HGP) (World Health Organization [WHO] grade sociated with depression of pancreatic exocrine function. II or III) from low-grade pancreatic neuroendocrine tumors (LGP) (WHO grade Material and methods: 1,458 volunteers aged 21 to 88 years (777 women) I). underwent an MRI of the pancreas, and organ fat content was quantified. Exo - Material and methods: A systematic literature search up to November 2017 crine pancreatic function was categorized by fecal elastase using defined cut- was performed to find relevant original studies. Retrieved hits were screened offs (normal function: >200 µg/g, n=1,319; moderate/manifest restriction: for inclusion and were evaluated with the revised tool for quality assessment <200µg/g, n=139). Statistical analysis to associate pancreatic fat content and for diagnostic accuracy studies (commonly known as QUADAS-2) by two re- fecal elastases includes a linear regression. Analyses were adjusted for known searchers. Pooled sensitivity, specificity, area under receiver operating charac - confounders for pancreatic steatosis, such as age sex and body mass index teristic (AUROC) curve, and other measures were calculated using random (BMI). effect models. Risk of heterogeneity was assessed for the appropriateness of Results: Overall mean (± standard deviation) of pancreatic fat content was meta-analysis. 7.50±3.78%. Significant higher levels were found in subjects with a restriction Results: Fifteen studies involving 866 patients were included. The statistical of pancreatic exocrine function (9.36±4.95%) compared to subjects with nor- heterogeneity was explained by imaging methods (CT/MR). The pooled AU- mal function (7.30±3.59%) (p<0.01). Linear regression analyses showed an ROC was 0.7494 for discrimination of HGP from LGP, with 0.632 (95% CI: inverse association between pancreatic fat and fecal elastase levels in the 0.567, 0.694) sensitivity and 0.688 (95% CI: 0.652, 0.724) specificity. whole study population [beta: -7.19 (standard error: 1.39); p<0.01] as well as in Conclusion: Hypo-enhancement in the arterial phase is a useful imaging fea- the subgroup of subjects with normal function [-4.26 (1.32); p<0.01]. Among ture to discriminate HGP from LGP. subjects with restriction in exocrine pancreatic function, a trend towards an inverse relation was detected [-1.28 (0.84); p<0.13]. Conclusion: There is an inverse relation between pancreatic fat content and SS 11.5 fecal elastases suggesting an association of pancreatic steatosis with a re- Pancreatic steatosis: quantitative analysis using MRI striction of pancreatic exocrine function. and correlation with visceral adiposity, diabetes mellitus, and chronic pancreatitis 1 1 1 2 1 T. Tirkes , T.A. Seltman , P.R. Territo , L. Li , S.A. Persohn , SS 11.7 1 3 1 1 M. Sankar , C.Y. Jeon , E.L. Fogel ; Indianapolis, IN/US, Intra-observer and inter-observer agreement on MRI 2 3 Houston, TX/US, Los Angeles, CA/US and MRCP features of chronic pancreatitis 1 2 3 4 2 T. Tirkes , A. Dasyam , Z. Shah , N. Takahashi , G. Tang , Purpose: To determine the association between the pancreatic steatosis (PS), 2 2 2 3 4 D. Yadav , K. Vipperla , P. Greer , D. Conwell , M. Topazian , visceral adiposity, diabetes mellitus and chronic pancreatitis (CP) using MRI. 1 1 2 E.L. Fogel ; Indianapolis, IN/US, Pittsburgh, PA/US, Material and methods: 118 patients were retrospectively categorized into 3 4 Columbus, OH/US, Rochester, MN/US normal (n=60), mild (n=21), moderate (n=27) and severe CP (n=10) groups based on clinical history and MRCP findings using the Cambridge classifica - Purpose: To determine intra- and inter-observer variability of MRI/MRCP for tion as the diagnostic standard. PS was calculated by measuring signal inten- evaluation of chronic pancreatitis (CP). sity on the axial water and fat-only fractions of the 2-point DIXON MR images. Material and methods: This ancillary study of the Consortium for the Study of Visceral and subcutaneous adipose tissues were separated and measured on Chronic Pancreatitis, Diabetes, and Pancreatic Cancer was performed at four axial MR image by manually outlining the anatomy using an image analyzer academic institutions. Anonymized MRI/MRCP scans of 32 patients, being software. evaluated for CP, were collected from one academic institution and reviewed Results: There was an association between the PS and distribution of the by three abdominal radiologists from other institutions. These radiologists abdominal adipose tissue, with the strongest correlation being with adiposity were asked to repeat the review two months later after scans were re-assigned within the visceral compartment (r=0.48). Higher visceral adiposity and PS a different case number. were also associated with CP; PS in the mild (24%), moderate (22%) and se- Results: Intra-observer agreement for Cambridge category varied from poor vere CP (21%) groups were significantly higher (p<0.0001) than the normal to excellent (kappa 0.33, 0.59, 0.81; concordance 47%, 69%, 84%). When group (15%). PS between the CP groups was similar (p=0.48). Smoking individual imaging findings were analyzed, intra-observer agreement was poor (p=0.003) and alcohol (p=0.02) were significant risk factors for CP. Patients to good for pancreatic atrophy (kappa 0.33, 0.66, 0.67; concordance 44%, who developed type 2 diabetes showed higher PS (p=0.01), higher visceral 53%, 66%), moderate to excellent for pancreatic duct (PD) caliber (kappa (p=0.01) and higher subcutaneous adiposity (p=0.002) as well as atrophy of 0.51, 0.65, 0.85; concordance 44%, 69%, 91%) and moderate to excellent for the tail (p=0.0009). There was a weak positive correlation between the PS and PD irregularity (kappa 0.45, 0.72, 0.89, concordance 66%, 69%, 89%). Inter- age in the normal group (r=0.33). observer for Cambridge category was moderate (kappa=0.52). Inter-observer Conclusion: Pancreatic steatosis quantitatively measured by the MRI showed agreement for individual imaging findings was poor for atrophy (kappa=0.32), a moderate positive correlation with visceral adiposity and was associated PD caliber (kappa=0.23), and PD irregularity (kappa=0.20). Patients were bal- with CP and type 2 diabetes. Prospective evaluation in well-phenotyped pa- anced in gender (50% female) with a median age of 50 years (IQR=37–58). tients is needed to confirm these results. Conclusion: There is mostly moderate, however, very variable intra- and inter- observer agreement for grading of CP using the Cambridge classification. A more robust, less subjective and more quantitative classification system, spe - cifically designed for MRI/MRCP seems necessary. This would allow improved categorization of CP patients for clinical and research purposes. S687 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 11.8 SS 11.10 MRI and MRCP features of chronic pancreatitis in MRI follow-up of intraductal papillary mucinous patients with or without pancreas divisum abnormality neoplasia: can we use a non-gadolinium-enhanced MRI and their relationship with patients’ age protocol? 1 2 2 2 M.L. Monti, G.A. Zamboni, L. Bertuzzo, R. Pozzi Mucelli, L. Bertuzzo , G.A. Zamboni , C. Fabris , R. Pozzi Mucelli , 2 1 2 G. Mansueto; Verona/IT G. Mansueto ; Santorso-VI/IT, Verona/IT Purpose: To compare the MRI and MRCP features of chronic pancreatitis (CP) Purpose: To assess whether non-contrast MRI can be safely used to follow up in patients with or without pancreas divisum (PD) and to determine their rela- patients with intraductal papillary mucinous neoplasia (IPMN). tionship with patient age. Material and methods: Institutional review board approval was waived for this Material and methods: We retrospectively reviewed 175 consecutive patients retrospective study. We evaluated 138 patients (48M, 90F, average age 64.5 (mean age 52 years) with symptomatic CP who underwent MRI and MRCP. 86 years) with an initial diagnosis of IPMN without signs of malignancy (128 had CP associated with pancreas divisum and 89 had CP without PD (biliary, branch-duct IPMN, 10 combined IPMN). All patients underwent at least two alcoholic, post-inflammatory, genetic, idiopathic). Both groups of patients consecutive MRI/MRCP with gadolinium in our center with an average follow- were stratified by age: <40 years (26 PD; 33 non-PD), 41–60 years (27 PD; 28 up of 12 months. For each patient, one reader evaluated, in separate sessions, non-PD), >60 years (33 PD; 32 non-PD). Two readers in consensus evaluated in the non-contrast and post-contrast scans: cyst size, walls/septa thickening, the scans for: morphological changes, signal intensity abnormalities, pancre- mural nodules, main pancreatic duct (MPD) dilation and contrast enhance- atic ducts changes, disease distribution. All MRI features were compared ment. across age groups using Chi-square tests. Results: We detected 376 cysts with a mean size of 10.9mm (5-29mm). Pa- Results: The pancreatic parenchyma was significantly more often T1-hypoin - tients had a median of 2 cysts each (range 1-20; mean 2.7). The MPD was di- tense in non-PD than in PD patients (p<0,0001). The degree of parenchymal lated in 22 patients, with a mean caliber of 4.9mm (4-9mm). In the first MRI thinning and irregularity of parenchymal margins was significantly higher in available, in the non-contrast scans, 356 (94.7%) lesions did not show suspi- non-PD patients in the groups aged <40 and 41–60 (p<0.0001) while no differ- cious signs. We detected walls/septa thickening in 4 cysts and endoluminal ence was observed between PD and non-PD patients aged >60 (p=ns). The defects in 16: 15 were classified as debris because of their dependent posi - main pancreatic duct (MPD) was significantly more often dilated in non-PD tion, 1 as mural nodule. After contrast administration, none of these showed patients, in all age groups (p=0.0031). Filling defects were more common in enhancement. At the next available follow-up MRI, in the non-contrast scans, non-PD patients (p<0.0007). A significant association was found between pa - 360 (95,7%) lesions did not show any changes, 11 (3%) showed a slight in- renchymal T1-hypointensity, MPD dilation and filling defects and increasing crease in size and 5 (1,3%) showed signs of progression. This was confirmed age in non-PD patients compared to PD patients in all age groups (p<0.001). in the post-contrast scans. Conclusion: Patients with PD show less severe morphological alterations than Conclusion: At follow-up, non-contrast MRI would have been enough to ex- patients affected by other forms of chronic pancreatitis; these morphological clude signs of malignant evolution in 98.7% cysts and 96.4% patients. alterations are significantly associated with increasing age. SS 11.9 Value of MRCP for the detection of intraductal papillary mucinous neoplasm of the pancreas V. Grossarth, J. Boddenberg, N. Ziayee, F. Verfürth, T. Lauenstein; Düsseldorf/DE Purpose: To assess the diagnostic value of MRCP for the detection of intra- ductal papillary neoplasm (IPMN) and to compare the results with endoscopic retrograde cholangiopancreatography (ERCP), endosonography and histopa- thology as the standard of reference. Material and methods: In this retrospective study, 98 patients with suspected IPMN were analyzed. All subjects underwent an MRI examination on a 1.5T scanner (Magnetom Avanto, Siemens) including 3D MRCP sequences. Two radiologists evaluated MRI data in a consensus mode regarding the presence of main duct (MD) or side branch (SB) IMPN. ERCP, endosonography and his- topathology were performed within two weeks of MRI and served as a stand- ard of reference. Results: Presence of IMPN was revealed in 80 of 98 patients. Overall sensitiv- ity and specificity of MRCP amounted to 97% and 62%, respectively. While sensitivity for the depiction of SB IPMN was as high as 97%, only 77% of MD IPMN was correctly identified by MRCP. Conclusion: MRCP is an accurate tool for the detection of IPMN in general. However, the diagnosis of MD IPMN should be verified in a clinical setting by further diagnostic tools since therapeutic options strongly depend on the cor- rect diagnosis. The only moderate specificity of MRCP may be related to the misdiagnosis of IPMN as other cystic pancreatic lesions. S688 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 The Liffey A SS 12.3 CTC in patients undergoing surgical management for Scientific Session SS 12 deeply infiltrating endometriosis CT colonography and colonic imaging S. Liddy, E. O‘Dwyer, R. Sarkar, A. O‘Neill, J. Feeney; Dublin/IE SS 12.1 Purpose: To determine the value of CTC in predicting the need for bowel re- Performance and evaluation in CTC screening section and stoma formation in patients undergoing surgical management for 1 2 1 A.E. Obaro , A. Plumb , R. Baldwin-Cleland , deeply infiltrating endometriosis. 1 2 1 1 C. Ugarte-Cano , S. Halligan , D. Burling ; Harrow/UK, Material and methods: A retrospective review of all patients who underwent London/UK surgical management of deeply infiltrating endometriosis at a tertiary referral centre over a 12-month period. CT colongraphy was correlated with intraop- Purpose: To implement a structured CTC training programme with individual- erative findings. Data recorded included patient demographics, medical his - ized feedback and assess radiologist attitudes to the programme and CTC tory, sites of disease involvement, the plane of contact between the bowel and quality metrics. the surrounding organs and/or endometrial deposits, the presence of luminal Material and methods: CTC reporting radiologists were randomised to either narrowing and the surgical procedure undertaken. A per-patient and per-lesion an intervention group (received one-day, 1:1 CTC training and feedback on analysis was performed using the Student’s t test, chi-square test and one- test performance) or a control group (no training or feedback). Both groups way analysis of variance. took four CTC assessments to assess detection of colorectal cancer (CRC) Results: Eighteen patients with a total of 25 sites of bowel involvement under- and polyps >5mm. Tests were administered at baseline, 1 month after work- went surgery in the 12-month period. The mean age was 39. Sites of bowel shop, 6 months and 12 months. CTC workload and views on CTC quality met- involvement included rectal (13/25), rectosigmoid (3/25), sigmoid (7/25) and rics were obtained via questionnaires. caecal (2/25). The mean plane of contact was 1.9 cm. Luminal narrowing was Results: 97 radiologists have been recruited and have a wide range of CTC present at 8/25 sites of bowel involvement. A longer plane of contact was as- reporting experience, from less than 50 cases (2 of 97, 2%) to over 3000 (10 of sociated with a need for bowel resection (2.6 vs 1.7 cm, p=0.04). Factors not 97, 10%). Most radiologists would find knowledge of their polyp detection rate associated with bowel resection included patient age (38 vs 39, p=0.44), mul- valuable or very valuable (92 of 97, 95%). 51 radiologists are randomised to the tifocal disease, site of bowel involvement and the presence of luminal narrow- intervention group. Of these, 28 have taken baseline test and attended the ing (2/7 vs 8/18, p=0.8). workshop. 71% (20 of 28) found the baseline test difficult, and all either agreed Conclusion: A longer segment of bowel involvement may be predictive of a (9 of 28, 32%) or strongly agreed (19 of 28, 68%) that the workshop provided need for bowel resection and stoma formation in patients undergoing surgical useful feedback regarding their baseline test performance as well as addi- management for deeply infiltrating endometriosis. tional learning opportunities (21 of 28, 75%, strongly agreed). 96% (27 of 28) felt motivated to independently improve their CTC performance. Conclusion: Early results demonstrate that a 1:1 CTC training workshop is SS 12.4 well received and motivates radiologists to improve their practice. Radiologists MR-enema (MR-e) and CTC in the diagnosis of recto- would find knowledge of their validated polyp detection rate very valuable. sigmoidal endometriosis: a preliminary comparative evaluation E. Biscaldi, F. Barra, C. Scala, V. Vellone, S. Ferrero; SS 12.2 Genova/IT Patient anxiety prior to a first CT examination to investigate abdominal malignancy: is CTC different from Purpose: To compare the accuracy of MR-e and CTC in detecting and evalu- standard abdominal CT? ating recto-sigmoid endometriosis. 1 2 2 1 D.J.M. Tolan , C. Roe , C.G.D. Clarke ; Harrogate/UK, Material and methods: We collected the patients following these inclusion Leeds/UK criteria: reproductive age, suspicion of deep pelvic endometriosis on the basis of symptoms and vaginal examination, presence of gastrointestinal symptoms, Purpose: This study evaluates patients undergoing CT for the first time and no previous radiological imaging. Patients underwent laparoscopy within 3 whether patients undergoing CTC have different levels of anxiety than those months from the diagnostic procedures. Radiological findings were compared having standard CT of the abdomen and pelvis (CTAP) when performed as a with surgery and histology. search for cancer. Results: Out of 85 women included, 52 had rectosigmoid endometriotic nod- Material and methods: Adult patients referred for either CTC or CTAP (single, ules (61.2%). The sensitivity, specificity, positive predictive value, negative dual or triple phase) were eligible for this pilot study. 60 consecutive eligible predictive value, likelihood ratio positive and likelihood ratio negative of MR-e patients were recruited (30 in each group). Participating patients completed and CTC in the diagnosis of recto-sigmoid endometriosis were 88.46% the state–trait anxiety inventory (STAI) questionnaire in the department prior to (95%C.I., 76.56%-95.65%), 90.0% (73.47%-97.89%), 93.88% (83.92%- CT scan. The study was granted health research authority ethics approval. 97.83%), 81.82% (67.75%-90.60%), 8.85 (3.01-26.0), 0.13 (0.06-0.27) and Results: Patients were recruited between March and October 2017. The CTC 92.31% (81.46%-97.86%), 93.75% (79.19%-99.23%), 96.0% (86.22%- group was older (CTC mean 77.7, range 63-90 vs. CTAP mean 56.6, range 25- 98.93%), 88.24% (74.44%-95.08%), 14.77 (3.85-56.64), 0.08 (0.03-0.21), re- 81, p=<.00001) and had more females (CTC n=23/30; 76.6% vs CTAP n=13/30; spectively. No significant difference was appreciated in the accuracy of the 43.3%, p=0.008). However, there was no significant difference in the baseline two techniques in the diagnosis of rectosigmoid endometriosis (p = 0.607; ‘T-anxiety’ measure (mean CTC 36.3 vs CTAP 34.6; p=0.42). The immediate McNemar’s test). CTC was more effective in evaluating the distance between pre-scan ‘S-anxiety’ was significantly higher for CTC (mean CTC 42.73 vs the nodule and the anal verge in all cases, it was less precise in evaluating the CTAP 33.73; p=0.00614). Patients having CTC have significantly higher anxi - depth of the lesions in the intestinal wall. CTC was more panoramic in whole ety levels before the scan from baseline levels (CTC mean T-anxiety 36.3, S- colon evaluation. anxiety 42.73, p=0.004296). Conclusion: CTC is competitive with MR-e in this diagnosis. CT software also Conclusion: Patients experience significant anxiety prior to CTC over those allows measurement of the distance of the nodule from the anal verge. CTC is attending for CTAP when investigating cancer. Further work is required to iden- probably still limited in evaluating the depth of penetration of the nodule in the tify the impact of anxiety on patient perception of care and effective approach- intestinal wall, due to the lack of contrast resolution. In the future, CTC may be es for managing patient anxiety in those having CTC. implemented with iodinated medium injection to boost its diagnostic yield. S689 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 12.5 SS 12.7 Incidence of clinically significant perforation at low- MRI parameters for prediction of sigmoid carcinoma in dose non-contrast CT and its value prior to same day patients with background of diverticular disease: a CTC following incomplete colonoscopy retrospective study 1 2 2 1 A. O‘Shea, M. Morrin, M. Lee, E. Thornton, T.É. Murray; E. Savoldi , A. Higginson , R. Beable , C. Ball ; 1 2 Dublin/IE Cosham/UK, Portsmouth/UK Purpose: Routine low-dose non-contrast CT of the abdomen and pelvis has Purpose: To retrospectively identify morphologic-specific findings at MRI in been recommended prior to same-day completion CTC to assess for occult the differentiation of mass-like diverticular disease from sigmoid carcinoma, in perforation at preceding incomplete colonoscopy, before insufflation at CTC. patients with background of diverticular disease and sigmoid stricture at CT. The aim of our study is to examine the incidence of clinically significant perfo - Material and methods: The cohort consisted of 37 patients with thickening/ ration at low-dose CT. We also examine the benefits of low-dose pre-scan in stricture of the sigmoid colon at CT, representing diverticular disease (n = 32) assessing adequacy of bowel preparation and identifying any other relevant or sigmoid carcinoma (n = 5). MRI studies were scored according to the pres- contraindications to CTC. ence or absence of potential discriminators. Patients underwent CT, T2- Material and methods: We conducted a retrospective review of all low-dose weighted and diffusion-weighted MRI. Sensitivity, specificity, positive predic - non-contrast CTs performed following failed colonoscopies over a one-year tive value (PPV), negative predictive value (NPV), and accuracy were calculated period (n=135). We also assessed the adequacy of bowel preparation on a and MRI were retrospectively reviewed using three different parameters: diffu- scale of 1-5, in order of increasingly adequate preparation. Incidentally noted sion restriction, loss of wall stratification and mucosal lesion. contraindications to CTC were also recorded. Results: MRI showed high NPV (93%) and specificity (96%) and PPV and Results: One perforation (which was extraperitoneal) was identified on low- sensitivity of 75% and 60%, respectively, with an overall accuracy of 91%. dose CTC. However, 6 patients were found to have contraindications to CTC Diffusion restriction had a sensitivity, specificity, and PPV of 100%, 96% and including acute diverticulitis and acute colitis. Overall, the bowel preparation 83%, respectively, for the diagnosis of cancer. Loss of wall stratification had a was felt to be adequate (≥ 3) in 96% percent of patients. 5 patients (0.04%) sensitivity, specificity, and PPV of 60%, 96%, and 75%, respectively, while the had their CTC postponed due to poor bowel preparation. presence of mucosal lesion had a sensitivity, specificity, and PPV of 80%, Conclusion: While the incidence of perforation identified at low-dose pre-scan 100%, and 100%. The analysis of the three parameters may change the MRI is low, there may be additional benefits to a low-dose non-contrast CT prior to accuracy at 97%. insufflation including assessment of adequacy of bowel preparation and iden - Conclusion: Diverticular disease is best differentiated from carcinoma by the tifying other contraindications to same-day CTC. analysis of three parameters: preservation of the wall layering, the absence of restricted diffusion and the absence of mucosal mass with low T2 signal. MRI improves the differentiation between sigmoid cancer and diverticulitis. These SS 12.6 encouraging results need to be confirmed in a larger study. 7 T MRI for fibrosis evaluation in a radiation-induced murine model of colitis 1 2 1 3 M. Zappa , S. Doblas , D. Cazals-Hatem , F. Milliat , SS 12.8 1 1 1 1 P. Garteiser , M. David , V. Vilgrain , B. Van Beers , Does the abdominal deep subcutaneous fat distribution 2 1 2 E. Ogier-Denis ; Clichy/FR, Paris/FR, influence complication rate and mortality after elective Fontenay-Aux Roses/FR surgery for colorectal cancer? E. Frostberg, Y. Manhoopi, M.R. Pedersen, H. Rahr, Purpose: Crohn‘s disease transmural bowel wall inflammation can lead to fi - S.R. Rafaelsen; Vejle/DK brosis causing luminal narrowing and fistula which are the main indications to surgery. The aim of our study was to evaluate MRI to distinguish submucosal Purpose: Intraabdominal visceral adipose tissue is associated with high risk of from transmural fibrosis using an original murine model of radiation-induced developing arteriosclerosis, and cardiovascular co-morbidity is associated colitis. with high complication and mortality rates after colorectal cancer surgery. Ac- Material and methods: Colitis was induced with localized single-dose radia- cording to the new literature, the deep subcutaneous adipose fat (dSAT) is like tion (27 Gy). We included an inflammation group of 24 rats with pathologic the intraabdominal visceral adipose tissue. The aim of this study was to evalu- features of severe acute inflammation associated with minimal submucosal fi - ate if the dSAT, measured by a regular CT scan, may be used as a predictor for brosis and a mixed group of 39 rats with pathological features of severe inflam - complications and mortality after colorectal cancer surgery. mation associated with transmural fibrosis, obtained two and twelve weeks Material and methods: Preoperative CT scans of 344 patients who under- after irradiation, respectively. Fat-suppressed T2- and T1-weighted, diffusion- went colorectal cancer surgery in the years 2010 and 2011 were examined weighted, magnetization transfer and perfusion using arterial spin labelling retrospectively. Measurements of the dSAT were collected. Clinical data re- (ASL) technique 7 T MRI was performed. garding patient characteristics, postoperative surgical and medical complica- Results: MRI showed significant differences between inflammation and mixed tions and survival data were retrieved. The radiological measurements were groups in normalized to muscle signal intensity on T2-weighted images (4.2 vs used as explanatory variables in fitted logistic and Cox regression models. 3.1, P< 0.0001) and T1-weighted images (1.4 vs 1.3, P=0.0003), ADC (2.17×10 Results: 325 patients were eligible for the study. The dSAT had a significant 3 -3 2 vs 1.69×10 mm /s, P<0.0001), magnetization transfer ratio (35 vs 42%, relationship with body mass index, and body mass index had a significant as - P<0.0001) and perfusion (60 vs 38 mL/min/100g, P=0.0009). Monoparametric sociation with surgical complications (OR: 1.09, p=0.01). Surgical complication analysis with the best area under the curve (AUC) to differentiate the two had a strong impact on medical complication (OR:5.9, p=0.0001). Union for groups included T2-weighted signal intensity, apparent diffusion coefficient international cancer control (UICC) stage, age and co-morbidity had a signifi - (ADC) and magnetization transfer ratio. The combination with the best global cant association with long-time survival. The radiological measurement of the predictive value (94%) included all parameters but T1-weighted parameter dSAT did not have an impact on surgical or medical complications, nor did it (AUC 0.94). have an association with long-time survival. Conclusion: MR multiparametric analysis was able to distinguish transmural Conclusion: Measuring the thickness of the dSAT in a preoperative setting from submucosal fibrosis in our radiation-induced model. This model could be does not have any impact on postoperative surgical or medical complications used to evaluate antifibrotic treatments currently under development. nor long-time survival in colorectal cancer patients treated with elective sur- gery. S690 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 Liffey Hall 1 SS 12.9 Model-based iterative reconstruction in ultralow-dose Scientific Session SS 13 CTC: diagnostic performance and influence on computer-assisted diagnosis Diffuse and chronic liver disease S. Barbieri, G. De Paoli Barbato, E. Raimondi, M. Bassi, A. Deledda, M. Tilli, R. Rizzati, M. Giganti, G. Benea; SS 13.1 Ferrara/IT Liver imaging findings in adult patients with Fontan Purpose: To compare the perfomance of standard dose (SD) and ultralow circulation dose (ULD) in CTC using hybrid-iterative reconstruction (HIR) and model- H.K. Yang, H.-J. Jang, R.M. Wald, S.-J. Yoo, T.K. Kim; based iterative reconstruction (MBIR), and to assess the effect of different re- Toronto, ON/CA construction algorithms on computer-aided diagnosis (CAD). Purpose: To describe liver imaging findings and complications on CT or MRI Material and methods: We evaluated 50 patients (31 males, 19 females, in adult patients with Fontan circulation and to investigate as to whether he- mean age 69.7) through CTC with SD (120kV, 50mAs 4.6±1.3mSv) and ULD patic imaging features show correlation with post-Fontan duration, age or se- (120kV, 10mAs, 0.95±0.2mSv); images were reconstructed with both HIR and rum hepatic function tests. MBIR. A commercial CAD system was applied complementary to all recon- Material and methods: Our institutional review board approved this retro- structions. Two independent radiologists reviewed all obtained images sepa- spective study with waiver of informed consent. Two blinded abdominal radi- rately and in random order to avoid recall biases. The two readers recorded ologists retrospectively reviewed the abdominal CT (n = 21) and MR (n = 16) colonic findings (polyps/cancer), extracolonic findings and CAD detections. images of 37 adults (mean age, 30 years; 14 men and 23 women) with Fontan Polyps (p) were classified by size (p<5mm, 5≤p<10mm, p≥10mm), shape (flat, circulation (mean post-Fontan duration, 23 years). Biochemical hepatic func- sessile, pedunculated) and location (rectum, sigmoid, descending, transverse, tion tests within 6 months of CT or MR examinations were used for analysis. ascending, cecum). Only relevant extracolonic findings (C-RADS: E4) were re - Results: Inhomogeneous, reticulated-mosaic pattern of hepatic parenchymal corded. CAD results were reported and significant findings were selected. enhancement, lobulated hepatic contour, blunt margin, and hypertrophy of the Results: 9 polyps were found in 8 patients from: six <5mm, two ≥5_<10mm left lateral section were seen in 86% (32/37), 96% (34/37), 95% (35/37) and and one ≥10mm; 8 were pedunculated and 1 flat 2 were in cecum, 3 ascend - 73% (27/37) of patients, respectively. Among the analyzed 25 arterial hyperen- ing, 2 transverse, 1 descending, 1 sigmoid tract; colonic-finding records were hancing nodules in 16 patients, 2 nodules (8%, 2/25) in 2 patients (13%, 2/16) superimposable for both radiologist with SD reconstructions and with ULD were hepatocellular carcinomas (HCCs) and the remaining 23 nodules in 14 MBIR reconstruction; 11 extracolonic findings E4 were recorded with SD, 2 patients were diagnosed as focal nodular hyperplasia (FNH)-like nodules. The with ULD (AAA, hiatal hernia). Colonic findings were confirmed by optical colo - diameter of inferior vena cava showed significant positive correlation with noscopy. post-Fontan duration, age and laboratory hepatic function test. Conclusion: CTC with ULD and MBIR reconstructions showed promising re- Conclusion: Imaging findings of adult patients with decades of Fontan circula - sults in terms of diagnostic performance and inter-reader reproducibility for tion were characterized by reticulated-mosaic pattern of parenchymal en- intra-colonic findings. CAD detection of significant findings was unaffected by hancement, lobulated contour, blunt margin, and hypertrophy of the left lateral acquisition protocol or reconstruction algorithm. section. Arterial-phase hyperenhancing FNH-like nodules are commonly seen, but HCC can infrequently occur. SS 12.10 Dynamic MR of the pelvic floor: impact of the SS 13.2 pubococcygeal line on the grading of pelvic floor Evaluation of two-point Dixon water-fat separation for descent gadoxetic acid-enhanced MRI-based estimation of liver 1 2 1 1 2 S. Picchia , M. Rengo , D. Bellini , M. Montesano , A. Laghi ; function 1 2 Latina/IT, Rome/IT M. Haimerl, S. Poelsterl, C. Stroszczynski, P. Wiggermann; Purpose: To evaluate the impact of three methods for drawing the pubococ- Regensburg/DE cygeal line (PCL) on the grading of pelvic floor descent with MR. Purpose: To evaluate the impact of intrahepatic fat on gadoxetic acid (Gd- Material and methods: Female patients with suspected pelvic floor dysfunc - EOB-DTPA)-enhanced T1 relaxometry for the estimation of liver function using tion were prospectively included in the study. All exams were acquired on a 1.5 T1-weighted volume-interpolated breath-hold examination (VIBE) sequence Tesla MR with patient in supine position. Rectum and vagina were filled with with Dixon water-fat separation. 200 ml and 50 ml of ultrasound gel, respectively. A distended bladder was re- Material and methods: 74 patients underwent Gd-EOB-DTPA-enhanced T1 quired. The acquisition protocol included turbo spin echo (TSE)-T2-weighted 13 13 relaxometry and C-methacetin breath test ( C-MBT) for the evaluation of sequences on the three planes for morphological evaluation. A steady-state liver function. Prior to (T1 ) and 20 minutes after the intravenous administra- pre sequence (True-FISP) in the midsagittal plane was acquired during straining tion of Gd-EOB-DTPA (T1 ), a T1-weighted VIBE sequence with Dixon water- post and defecation, acquiring 1 image/sec. The anterior point of all PCL was the fat separation and multiple flip angles (1°, 7°, and 14°) was used to generate pubic symphysis, while posterior points were the tip of the coccyx (PCLtip), both T1 maps (T1_in) of the in-phase component without fat separation and T1 the sacrococcygeal joint (PCLsc) and the last coccygeal joint (PCLcc). The maps merely based on the water component (T1_W). The reduction rates of T1 grade of pelvic floor descent was measured with the M-line according to the relaxation time (rrT1) between pre- and post-contrast images were evaluated HMO system, traced perpendicularly from the PCL to the posterior end of the 13 for T1_in (rrT1_in) and T1_W (rrT1_W). C-MBT values were correlated with T1 H-line. Measurements of M-line were compared as differences of absolute and rrT1. post measures and differences of grading. Results: A constant significant decrease of rrT1 with progressive reduction of Results: The final population consisted of 29 consecutive patients (mean age liver function could be shown for both T1_in (p < 0.001) and T1_W (p < 0.001). 56.07yy±10.52). M-line measured using PCL tip was significantly smaller than 13 When patients are subdivided into 3 different categories of C-MBT readouts, PCLsc or PCLcc in all measurements. The grading of pelvic floor relaxation, the groups can be significantly differentiated by their rrT1_in and rrT1_W val - using PCLtip, was underestimated in 13 patients (44%) at rest, in 19 patients 13 ues (p<0.005). Simple regression model showed a log-linear correlation of C- (65%) during straining and in 16 patients (55%). MBT values with T1_in (r = 0.566; p<0.001), T1_W (r = 0.568; p<0.001), post post Conclusion: A significant underestimation of the pelvic floor descent was ob - rrT1_in (r = 0.726; p < 0.001) and rrT1_W (r = 0.766; p<0.001). served when the M-line was measured using the PCLtip as reference. The use 13 Conclusion: Liver function as determined using C-MBT can be estimated of PCLsc or PCLcc should be used to quantify the pelvic floor descent. from Gd-EOB-DTPA-enhanced MR-relaxometry more accurately when intra- hepatic fat is taken into account. Here, T1_W maps are significantly superior to T1_in maps without separation of fat. S691 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 13.3 SS 13.5 Disodium gadoxetate-enhanced liver MRI in cirrhotic Assessment of liver function in patients with chronic patients: can clinical parameters predict suboptimal liver diseases using different parameters of gadoxetic hepatobiliary phase enhancement? acid-enhanced hepatobiliary phase imaging: a S.P. Walker, D. Wilson, H. Ramotar, C.R. Smith, R. Albazaz, comparison study M.B. Sheridan, J.A. Guthrie; Leeds/UK L.B. Beer, N. Bastati-Huber, S. Pötter-Lang, D.P. Stoyanova, M. Elmer, D. Tamandl, A. Ba-Ssalamah; Vienna/AT Purpose: The potential added value of disodium gadoxetate (Gd-EOB-DTPA) is its active uptake by hepatocytes to provide a hepatobiliary phase (HBP) aid- Purpose: To evaluate the correlations between different hepatobiliary phase ing detection and characterisation of focal liver lesions. Predicting suboptimal (HBP) image scores using gadoxetic-acid (GA) uptake and the albumin-biliru- HBP enhancement would be advantageous so as to select a different contrast bin (ALBI) grading system, in patients with chronic liver disease. medium saving expense and room time. This study evaluates the relationship Material and methods: This retrospective institutional review board approved of clinical parameters to HBP enhancement using Gd-EOB-DTPA in cirrhotic study included 220 patients (139 male, 81 female, mean age 54.9 ±14.3 years) patients. with chronic liver disease who had undergone GA-enhanced MRI of the liver Material and methods: Cirrhotic patients undergoing Gd-EOB-DTPA MRI be- between 2011 and 2015. Relative liver enhancement (RLE), hepatic uptake tween January 2016 and December 2016 were retrospectively reviewed. index (HUI), contrast uptake index (CUI), and liver-to-spleen contrast ratio Based on region-of-interest sampling, the liver signal intensity was calculated (LSC) were calculated using unenhanced and GA-enhanced HPB images, 20 using the spleen as a reference. The liver spleen contrast ratio (LSCR) was minutes after administration of the contrast material. ALBI grading was used then calculated and compared with basic patient characteristics and labora- as a clinical score for correlation with disease severity. Correlation coefficients tory results. between different factors were calculated using Pearson‘s correlation coeffi - Results: 127 patients were identified with an average HBP time of 19 minutes cient. 31 seconds. Twelve patients were excluded due to incomplete availability of Results: The highest correlation coefficients were observed between RLE-CEI clinical parameters. Multiple logistic regression analysis revealed that bilirubin (R=0.904, p<0.001), RLE-LSC (R = 0.782, p<0.001), and CUI–LSC (R=0.787, (p=0.04), albumin (p=0.01), PT (p=0.02) and platelets (p=0.04) were factors p<0.001). Correlation strength between HUI and RLE, CEI, and LSC was predictive of HBP enhancement. These clinical parameters were also signifi - (R=0.564, R = 0.535, R = 0.434, respectively, p<0.001 for all). R values be- cant with a simple logistic regression analysis, except for platelets (p=0.86). tween the ALBI score and RLE, CUI, HUI, and LSC were -0.481. -0.442, -0.335, Conclusion: HBP enhancement in cirrhotic patients correlates with a number -0.416, respectively, indicating a moderate correlation. of clinical biomarkers, the strongest being albumin. Prospective identification Conclusion: A strong correlation between different MR-derived parameters of suboptimal HBP enhancement using such biomarkers could avoid the non- using HBP imaging and the ALBI score, except for the HUI, was observed. beneficial use of Gd-EOB-DTPA in cirrhotic patients thus saving money and These data indicate that, except for volume-based scores, such as the HUI, resources in favour of alternative contrast medium. other MR-derived parameters are interchangeable for the evaluation of diffuse liver disease. SS 13.4 Assessing liver function: diagnostic efficacy of SS 13.6 parenchymal enhancement and liver volume ratio during Comparison of hepatic R2* and fat fraction obtained by hepatobiliary phase of gadolinium-ethoxybenzyl- pixelwise map methods to MRQuantif software results diethylenetriamine pentaacetic acid-enhanced MRI A. Paisant, C. Chargeboeuf, J. Jouan, H. Saint-Jalmes, studies E. Bardou-Jacquet, Y. Gandon; Rennes/FR A. Pecorelli, E.B. Orsini, C. Talei Franzesi, S. Famularo, Purpose: The aim was to compare the results of MRQuantif freeware using a L. Gianotti, D. Ippolito; Monza/IT generic 2D multi-echo gradient echo sequence (ME-GRE), previously validated Purpose: To assess whether gadolinium-ethoxybenzyl-diethylenetriamine against biopsy, to constructor sequences providing R2* and fat fraction (FF) pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI study is useful to determine maps. liver function in comparison to Child-Pugh (CP), model for end-stage liver dis- Material and methods: After institutional review board (IRB), also known as ease (MELD) and biochemical test. an independent ethics committee (IEC), ethical review board (ERB), or re- Material and methods: All Gd-EOB-DTPA-enhanced MRI studies performed, searchIRB approval and written informed consent, on a 3.5-year period, we between May 2011 and September 2017, to characterize focal liver lesion explored 619 patients for hepatic iron or fat quantification. For each we ob - were retrospectively reviewed. Patients were divided into study and control tained a generic 2D ME-GRE with TE multiple of 1.2 ms and a constructor 3D group according to the presence/absence of liver cirrhosis. Signal intensity ME-GRE providing a R2* or T2* and a FF maps, either Ideal-IQ (n=113) on a was calculated as the rate of liver-to-muscle ratio on contrast-enhanced T1- 1.5T MR450 (General Electric), 3D-Vibe-Dixon WIP (n=196) on a 3T Verio (Sie- GE sequences in portal (SI-POR) and hepatobiliary phase (SI-HEP) and then mens) or mDixon-Quant (n=310) on a 3T Ingenia (Philips). We then compared normalized for liver volume (SI-POR/LV and SI-HEP/LV). (correlation and Bland-Altman) R2* and FF determined by MRQuantif to the Results: A total of 303 Gd-EOB-DTPA-enhanced MRI studies, performed in value of the two maps, from the mean of 3 region of interests (ROI)s placed at 221 consecutive patients, were included. Cirrhosis was present in 191 (63.0%) the same location on the right liver. Discordant data were specifically analyzed MRI studies. SI-HEP was significantly lower in cirrhotic than in non-cirrhotic by an expert. patients (0.55±0.29 vs 0.66±0.40, p=0.004), while no difference was found in Results: After exclusion of map reconstruction miscalculation errors (FF>50%, SI-POR. SI-HEP progressively decreased from CP-A group to CP-C (0.59±0.28 n=6), R2* and FF correlations of MRQuantif to 1.5T Ideal-IQ were r²=0.6 and to 0.25±0.19, p<0.0001) and from MELD≤10 to MELD 19-24 (0.58±0.30 to 0.6, to 3T Vibe-Dixon were r²=0.76 and 0.31 or to 3T mDixon-Quant were 0.54±0.49, p=0.773). SI-HEP had a good performance in distinguishing CP-A r²=0.94 and 0.87, respectively. Correlations increased greatly when excluding from CP-B/C patients (AUC=0.75; 95%CI=0.66-0.83). Among biochemical pa- high liver iron overload with still clinically significant map results errors. 3T R2* rameters, a moderate correlation was found between SI-HEP and total biliru- maps were not accurate above 200 µmol of iron/g liver. bin (R=-0.324), GOT (R=-0318) and albumin (R= 0.320). Comparable results Conclusion: R2* and FF maps are less robust than the ROI analysis proposed were observed when SI-HEP was normalized for liver volume. by MRQuantif.org freeware. Conclusion: SI-HEP of Gd-EOB-DTPA-enhanced MRI studies can be used to effectively evaluate liver function. In clinical practice, this imaging technique could be performed both to correctly characterize liver lesions and to assess the severity of liver disease especially in the perspective of surgical treatment. S692 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 13.7 SS 13.9 MRI hepatic fat quantification using various methods, at Liver surface nodularity quantification from routine CT 1.5 and 3T, in comparison to quantitative images for the detection of clinically significant portal histomorphometry hypertension 1 1 1 1 1 1 2 1 1 A. Paisant , J. Jouan , E. Bardou-Jacquet , E. Bannier , R. Sartoris , P.-E. Rautou , L. Elkrief , V. Vilgrain , M. Ronot ; 2 2 1 1 2 1 2 J. Boursier , C. Aubé , Y. Gandon ; Rennes/FR, Angers/FR Clichy/FR, Geneva/CH Purpose: To compare the results of different methods of quantification by MRI Purpose: To determine whether quantification of liver surface nodularity (LSN), of the hepatic fat fraction (FF), at 1.5T and 3T, with the histological, semi- a computer-based marker derived from routine CT images, can estimate the quantitative and especially quantitative measurement of steatosis. presence of clinically significant portal hypertension (CSPH) in patients with Material and methods: 169 patients were prospectively included in 2 centers, cirrhosis, and to compare LSN with other existing non-invasive tests. one using a 1.5T Signa (GEMS), the other using a 3T Verio (Siemens) MRI. A 2D Material and methods: This retrospective study included 189 patients with gradient echo multi-echo sequence (2D-ME-GRE), designed for hepatic fat cirrhosis who underwent hepatic venous pressure gradient (HVPG) and ab- and iron assessment, was used to quantify FF. MRquantif software calculated dominal CT, including 102 (54%) with CSPH. LSN was measured on CT and Dixon method (dual-echo, triple-echo or with T2* correction) and „complex“ other available non-invasive tests were performed, namely liver and spleen method taking into account a 3-peak spectral modeling of the fat. All patients volumes, liver-to-spleen volume ratio, platelet/spleen diameter, Iranmanesh had a liver biopsy with semi-quantitative assessment and a quantitative meas- score, aspartate aminotransferase/platelet ratio index (APRI), Fibrosis-4 (FIB- urement of FF by determining the percentage of surface of the fat vacuoles. 4). Technical success rate and measurement time were calculated. Area under Analyses used were Spearman test and Bland-Altman comparison. receiver operating characteristic (AUROC) and DeLong test were used. Results Results: MRI and quantitative FF were well correlated: Dixon dual-echo were validated in an independent external cohort of 78 patients with biopsy- (ρs=0.95, p<0.001 at 1.5T; ρs=0.87, p<0.001 at 3T), triple-echo (ρs=0.95, proven cirrhosis. p<0.001 at 1.5T; ρs=0.93, p<0.001 at 3T), corrected by T2* (ρs=0.94, p<0.001 Results: Technical success rate of LSN was 93%; measurement time was at 1.5T ; ρs=0.94, p<0.001 at 3T), complex (ρs=0.94, p<0.001 at 1.5T; ρs=0.94, 113±36.3 seconds. LSN quantification was correlated to the HVPG (r=0.75; p<0.001 at 3T). The biases calculated by the Bland-Altman method were p<.001). Patients with CSPH had a significantly higher LSN score than those closed to 0. Correlation of semi-quantitative histological assessment to MRI without (3.22±0.59 vs. 2.44±0.34; p<.001). A cut-off value of 2.76 had a posi- FF and also to quantitative histological FF was weaker. tive predictive value of 88% for CSPH. AUROC of LSN for the estimation of Conclusion: Using a 2D-ME-GRE sequence, MRQuantif freeware provides a CSPH was 0.88. This was significantly higher than that of other available non- good evaluation of FF compared to a really quantitative gold standard at 1.5T invasive tests (all p<.001). Cut-off value 2.76 was validated in the independent and 3T. MRI FF assessment is a better gold standard than the semi-quantita- external cohort. tive histological grading. Conclusion: Non-invasive CT-based LSN quantification rapidly and accurate - ly detects CSPH. This strategy could be particularly useful to evaluate portal hypertension before liver surgery. SS 13.8 Liver MR relaxometry: basic rules for clinical application and reference values when analyzing T1 and T2* maps SS 13.10 V.C. Obmann, N. Mertineit, C. Marx, A. Berzigotti, Modern MRI in the diagnostic workup of cystic fibrosis- J. Heverhagen, A. Christe, A.T. Huber; Bern/CH associated liver disease S. Pötter-Lang, K. Staufer, P. Baltzer, D. Tamandl, Purpose: To determine T1 mapping reference values of the liver at 3T with N. Bastati-Huber, D. Muin, L. Kazemi-Shirazi, respect of hepatic fat, fibrosis and iron content, as well as susceptibility-in - A. Ba-Ssalamah; Vienna/AT duced liver effects from the adjacent lung parenchyma. Material and methods: 143 patients without solid liver lesions >1cm on con- Purpose: To identify independent imaging features and establish a diagnostic trast-enhanced abdominal CT scans underwent multiparametric MR imaging algorithm for early diagnosis of cystic fibrosis-associated liver disease (CFLD) on a Siemens Verio 3T magnet including T1-/T2*-mapping, proton density fat in CF patients as compared with a control group (CG) on gadoxetic acid-en- fraction (PDFF) quantification and MR elastography. The population was di - hanced MRI. vided into negative patients without fibrosis (shear modulus <2.8 kPa) and Material and methods: A total of 90 adult patients were enrolled, 50 with CF steatosis (PDFF<10%) and positive patients with fibrosis and/or steatosis. T1 and 40 of the CG. The CF group was divided into two subgroups, the retro- values were compared between liver segments and groups. Multivariate analy- spective or test subgroup (n = 33) and the prospective or validation subgroup sis was used to statistically correct T1 for T2*. Mann-Whitney U test between (n = 17). The CG (patients with normal liver enzymes, but no more than benign segments and groups was performed. focal liver lesions) was divided also accordingly (27:13). MRI variables includ- Results: In negative patients, T2*-time was a significant confounder for T1- ing quantitative and qualitative parameters were used to distinguish those with time (p<1^15), while age and sex (p=0.722, 0.687) were not. T1 and T2 relaxa- CFLD from the CG using the clinical criteria by Debray et al. Fifteen qualitative tion times were shorter in lung-adjacent liver segments as compared with non- single-lesion CF descriptors were defined. Two readers independently evalu - lung-adjacent segments (mean of differences 33 ms, p<0.001 for T1). Normal ated the images. Univariate statistical analysis was performed to obtain the T1 in reference patients’ non-lung-adjacent segments was 780±83 ms. In significant imaging features that differentiate CF patients from the CG. Subse - positive patients with steatosis and fibrosis, T1 was significantly higher quently, a multivariate classification analysis using the chi-squared automatic (p=0.011 or less). When correcting T1 for T2*, this discrimination between interaction detector (CHAID) methodology was performed to identify the most positive and negative patients could be improved (p<0.001 for all compari- important descriptors. Diagnostic performance was assessed by receiver-op- sons). erating characteristic (ROC) analysis. Conclusion: When analyzing T1 maps in the liver at 3T, non-lung-adjacent Results: Three independent imaging descriptors distinguished CFLD from CG: segments should be measured due to susceptibility effects of the adjacent a) the presence of altered gallbladder morphology (GBAM); b) periportal track- lung. Liver iron content as indicated by T2*-shortening is a major confounder ing (PPT) and c) periportal fat deposition (PPFD). Prospective validation of the of T1 time. classification algorithm demonstrated a sensitivity of 94.1% and specificity of 84.6% for discriminating CFLD from the control cohort. Conclusion: We identified 3 independent imaging features that could poten - tially diagnose early-stage CFLD on unenhanced MRI. S693 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 Wicklow Hall 2A SS 14.4 Virtual unenhanced imaging of the liver derived from Scientific Session SS 14 160mm rapid-switching dual-energy CT: comparison of Technical advances in abdominal imaging image quality and solid liver lesion conspicuity with true unenhanced images 1 1 1 1 2 M. Lacroix , S. Mulé , E. Hérin , F. Pigneur , P. Richard , SS 14.1 1 1 1 1 B. Zegai , L. Baranes , M. Djabbari , F. Brunetti , withdrawn by the authors 1 1 1 1 N. De Angelis , A. Laurent , A. Rahmouni , A. Luciani ; 1 2 Creteil/FR, Buc/FR SS 14.2 Purpose: To compare virtual unenhanced images derived from arterial (VUEart) Diagnostic accuracy of whole body and portal-venous phases (VUEport) with native unenhanced (NU) acquisitions 18F-fluorodeoxyglucose positron emission tomography- in patients with suspected solid liver tumors (SLT) imaged on a rapid-switching MR in HCCs: comparison of local and metastatic single-source 160mm dual-energy CT (rsDECT). staging with chest-abdomen-pelvic CT and liver MRI Material and methods: 73 patients with 52 SLT who underwent a multiphasic 1 1 1 1 M. Vermersch , B. Emsen , S. Mulé , J. Chalaye , abdominal rsDECT were included in this institutional review board approved 2 1 1 1 1 A. Monnet , E. Hérin , F. Pigneur , L. Baranes , H. Regnault , retrospective study. Both NU (120kVp, 40mm, Pitch 1.375, rotation 0.7s, 1 1 1 1 1 G. Amaddeo , A. Laurent , A. Rahmouni , E. Itti , A. Luciani ; ASIRv 40%), VUEart and VUE port (80 kVp-140 kVp; 80mm collimation, pitch 1 2 Creteil/FR, Saint Denis/FR 0.992, rotation time 0.8s, ASIRv 50%) were reconstructed from raw data. Both Purpose: To assess the diagnostic accuracy and therapeutic management qualitative detectability of SLT on a 4-point scale and quantitative analyses of impact of whole body 18F-fluorodeoxyglucose positron emission tomogra - liver and erector-spinae muscles mean+SD densities, and lesion-to-liver con- phy-MRI (WB-18FDG-PET-MR) for local and whole body staging of HCCs. trast-to-noise-ratio (CNR) were compared on NU, VUEart and VUEport Material and methods: 36 consecutive patients followed for HCC were in- (Kruskal–Wallis, Dunn, Kolmogorov–Smirnov). cluded in this retrospective study. All patients underwent WB-18FDG-PET-MR Results: The mean liver density was significantly lower on VUEart and VUE - (Biograph mMR) which included whole body sequences (simultaneous multi- port than on NU (respectively, 56.7±6.7 and 57.9±6 vs 59.6±7.5; p<0.01), but slice with blipped Controlled Aliasing in Parallel Imaging Results in Higher Ac- was similar on VUEart and VUEport (p=0.09). The mean muscle density was celeration sequence design (SMS diffusion-weighted imaging (DWI)), unen- significantly lower on VUEart and VUEport than on NU (respectively, 47.1±5.4 hanced and post contrast 3DT1 Dixon VIBE) simultaneously acquired with PET and 48.4±5.3 vs 51.8±5.2; p<0.0001), but was similar on VUEart and VUEport data (25 minutes acquisition overall), and dedicated liver sequences (SMS in- (p=0.5). The lesion-to-liver CNR was significantly higher on VUEart and VUE - travoxel incoherent motion (IVIM) DWI, T2, multiphasic 3DT1 VIBE). The ac- port than on NU (respectively, 1.7±1 and 1.6±1.1 vs 0.9±0.6; p<0.001), but was curacy of WB-18FDG-PET-MR for HCC detection, local—biliary, vascular in- similar on VUEart and VUEport (p>0.9). Lesion conspicuity was significantly volvement—regional N staging, and distant staging was compared to that higher on VUEport images than on NU (p=0.004). observed with chest-abdomen-pelvis (CAP) CT and liver MRI, with pathology, Conclusion: rsDECT-derived VUE improves lesion-to-liver CNR and image follow-up or consensus interpretation of all available data as the reference quality and could replace true unenhanced acquisitions in this oncology set- standard. ting. Results: Overall, 55 HCC and 7 metastatic sites were documented in 33 pa- tients. Sensitivity, specificity, positive predictive value (PPV) and negative pre - SS 14.5 dictive value (NPV) of WB-18FDG-PET-MR were 100% for liver HCC detection, Accuracy of multi-parametric MRI in the diagnosis and local and regional staging. The sensitivity of WB-18FDG-PET-MR was 100%, grading of intestinal acute graft-versus-host disease specificity 97%, PPV 83%, NPV 100% and accuracy 97% for detection of after allogenic bone marrow transplantation metastatic patients. WB-18FDG-PET-MR led to changes in therapeutic man- D. Vitale, F. Maccioni, V. Buonocore, D. Bencardino, agement in 14% of patients (5/36), 2 by detection of additional HCC and 3 by M. Lopez, A.P. Iori, C. Catalano; Rome/IT detection of additional metastatic sites. Purpose: The diagnosis of intestinal acute graft-versus-host disease (a- Conclusion: One-stop-shop WB-18FDG-PET-MR improves local and distant GVHD) after allogenic bone marrow transplantation is based on clinical symp- HCC tumor staging compared to CAP CT and liver MRI performed separately, toms, CT findings and biopsies. We assessed MRI diagnostic accuracy. with significant impact on patient treatment management. Material and methods: We performed a prospective study on 35 patients (range 9-69 years) with clinical suspicion of intestinal a-GvHD, who underwent SS 14.3 clinical-endoscopic, histological and multi-parametric MRI evaluation. Inclu- withdrawn by the authors sion criteria: intestinal symptoms within 100 days from transplantation, <2 weeks between MRI and clinical-endoscopic evaluation. The following MRI features were evaluated: small and large bowel wall thickening, T2 signal and gadolinium enhancement of the intestinal wall, mesenteric lymph nodes, peri- toneal effusion, comb-sign and restricted diffusion. Histology, clinical-endo- scopic data and follow-up were considered as gold standard for diagnosis and staging. Sensibility, specificity, accuracy and Cohen’s kappa were calculated. Results: In 21/35 (60%), a-GvHD was confirmed at histology and follow-up. In true-positive patients, MRI showed significant continuous wall thickening in 76.2%, stratified wall contrast enhancement in 90%, comb-sign in 76%, in - creased number of mesenteric lymph nodes in 19%, and free intra-peritoneal fluid in 57.2%. A significant correlation was found between the number of pathological intestinal segments and clinical grade of a-GvHD (r=0.54, p=0.009). The distal ileum was the segment most frequently involved (85.7%). MRI sensitivity was 90.5%, specificity 78%, the PPV 86%, NPV 84% and ac - curacy 85%. Furthermore, MRI detected early disease in 82% of patients with- out intestinal symptoms. Conclusion: In patients with intestinal a-GvHD, MRI can diagnose and grade the disease with high accuracy, in both early and more advanced or severe stages. S694 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 14.6 SS 14.8 How low can we go? Assessment of the lowest CT Transient arterial phase respiratory motion-related enterography dose achievable with iterative artifact in liver MR imaging with gadoxetic acid vs. reconstruction extracellular gadolinium: a within-patient cohort 1 1 2 3 M. Twomey , H. Javed , D. Mondal , S. Moghe , comparative study 1 1 1 1 J. Sammon , P. Rogalla , L. Guimaraes ; Toronto, ON/CA, E. Belmonte, A. Darnell, J. Rimola, A. Forner, V. Sapena, 2 3 Oxford/UK, Vernon Hills, IL/US J. Bruix, C. Ayuso; Barcelona/ES Purpose: To determine the maximum dose reduction achievable without sac- Purpose: To compare the frequency and severity of transient arterial phase rificing observer performance for the detection of Crohn’s disease (CD) lesions (TAP) respiratory motion-related artifact following gadoxetic acid (EOB) and in CT enterography (CTE) using filtered back projection (FBP), hybrid iterative extracellular gadolinium (Ec-Gd) in a cohort of cirrhotic patients. reconstruction (H-IR) and a new model-based IR (MB-IR) algorithm. Material and methods: Eighty-two cirrhotic patients from prospective series Material and methods: 50 CTE studies performed in a 320 row scanner for underwent liver MR with EC-Gd and EOB within 1 month. Two readers, blinded known/suspicion of CD between July 2016 and July 2017 were included. A to all data, assigned a respiratory motion-related sore (0 [none], 1 [mild, no validated noise insertion tool was utilized to simulate 75%, 50% and 25% of impact on reader confidence interpretation], 2 [moderate, causing decrease in original dose. Datasets were reconstructed with FBP, H-IR and 2 types of MB- reader confidence interpretation], and 3 [sever, non-diagnostic]) for non-en - IR (IR1 and IR2). Two fellowship-trained abdominal radiologists independently hanced, arterial and portal venous phase. Scores for each phase were aver- reported intestinal findings and assessed image quality (IQ), noise and sharp - aged, and definitions for TAP, severe TAP and post-contrast motion-related ness using 5-point scales. Assessment of the full-dose dataset/endoscopic artifacts were established. The frequency of motion-related artifacts was com- data by an experienced abdominal radiologist served as reference standard. pared for each pair of examinations. Results: Fifty patients with 48 CD lesions were included. Mean effective dose Results: The mean motion scores for EC-Gd and EOB for non-enhanced of the full-dose datasets was 3.9 mSv. At 50% & 75% dose, sensitivity was phase were 0.68 vs 0.73 (p=0.3), for arterial phase 0.91 vs. 1.26 (p<0.0001), 98% for both readers and all reconstruction methods. At 25% dose sensitivi- and for portal venous phase 0.73 vs. 0.87 (p=0.04). The frequency of TAP ob- ties of FBP, H-IR, MB-IR were 88%, 92%, 88% for Reader 1 and 84%, 96%, served on EOB was superior to EC-Gd (19.5% vs. 6%; p=0.013), but there 89% for Reader 2. Agreement was almost perfect (k=0.82). IQ of FBP was were no differences in frequencies for severe TAP (p>0.99) between both con- unacceptable and worse than H-IR and MB-IR (p<0.05) at all dose levels. trasts agents. EOB was associated with a higher frequency of post-contrast Mean H-IR and MB-IR IQ scores were <3 (acceptable) at 50% and 75% dose. motion-related artifacts (7.3% vs. 0%; p=0.03). No significant difference was seen between H-IR and MB-IR1/2 (p>0.1357). Conclusion: TAP respiratory motion-related artifact is more frequently ob- MB-IR1 images were the sharpest, while MB-IR2 were less noisy (p<0.05). served on liver MRI with EOB than with EC-Gd and it may affect 19.5% of the Conclusion: Low-dose and ultralow-dose CTE can be performed with H-IR patients. However, the rate of severe TAP was very low and similar for both and MB-IR with acceptable image quality without compromising observer per- contrast agents. formance, providing knowledge facilitating confident implementation of lower dose CTE protocols using multiple CT platforms and reducing radiation expo- SS 14.9 sure in this patient population. withdrawn by the authors SS 14.7 SS 14.10 The effect of gadoxetic acid on image quality of multiple Assessment of whole-body fat using MRI: a pilot study hepatic arterial phase imaging: comparison study with 1 2 2 2 J.-P. Kühn , N. Friedrich , M.-L. Kromrey , J. Effler , gadoteric acid-enhanced MRI 1 1 3 1 R.-T. Hoffmann , M. Laniado , R. Laqua ; Dresden/DE, K.C. Sim, B.J. Park, H. Na Yeon, M.J. Kim, D.J. Sung, 2 3 Greifswald/DE, Bern/CH S.B. Cho; Seoul/KR Purpose: To develop an algorithm to assess fat compartments of the whole Purpose: To evaluate the effect of gadoxetic acid on image degradation of body using confounder-corrected chemical shift-encoded magnetic reso- single-breath-hold, triple (first, second, and third) arterial phase acquisition of nance imaging (CSI-MRI). the liver compared with those obtained with gadoteric acid. Material and methods: Seventy-three volunteers aged 29 to 80 years (29 fe- Material and methods: A total of 211 consecutive MR examinations were male) underwent a whole-body MRI including three-echo chemical shift-en- evaluated including triple arterial phase acquisition using 4D eTHRIVE Keyhole coded sequences acquired in 5 steps. After confounder correction and stitch- technique. 108 examinations were performed with gadoxetic acid and 103 ex- ing of steps, the whole-body proton-density fat fraction was acquired. Based aminations were performed with gadoteric acid. Two radiologists evaluated on tissue fat content, we developed a post-processing for segmentation of fat image qualities of the precontrast and triple arterial phases of both groups in a compartments as well as for tissue fat quantification. Using this algorithm, two blinded fashion. Image quality on each phase was assessed in regard to image observers measured total body fat volume (TAT), subcutaneous fat volume artifacts, sharpness of intrahepatic vessel, sharpness of liver edge, and overall (SAT), visceral fat volume (VAT), and liver fat content (LFC). Robustness of the image quality. algorithm was tested by evaluation of inter- and intraobserver variability using Results: Gadoxetic acid group shows better image score in the precontrast a Bland-Altman analysis. In addition, data were correlated with body mass in- images, despite the compromised conditions such as cirrhosis, more ascites, dex (BMI) and age using Spearmen correlation coefficients. and variable comorbid malignancies that could make more degraded image Results: TAT, SAT, FAT and LFC were successfully measured in all volunteers. quality (p < 0.05). Image quality of triple arterial phases of gadoxetic acid-en- There were no differences between two readings (mean bias (%)±standard hanced liver MRI was significantly degraded than those of gadoteric acid for deviation; TAT/SAT/FAT/LFC: -0.1±0.3/0.4±1.0/0.2±0.4/0.4±1.0) and both ob- artifact, sharpness of intrahepatic vessel, sharpness of liver edge, and overall servers (TAT/SAT/FAT/LFC: 0.0±0.3/0.9±1.4/0.4±0.8/0.9±1.6). TAT showed the quality in each arterial phases (p < 0.05). Overall image quality was gradually highest correlation with BMI (r=0.88, p<0.01), followed by SAT (r=0.83, p<0.01), improved from first to third arterial phase (p < 0.003). Third arterial phase VAT (r=0.76, p<0.01), and LFC (r=0.61, p<0.01). The order was not affected by showed best image quality in both groups. Inter-rater reliability was high (p < sex; however, the correlation coefficients were generally higher in men than in 0.001). women. Conclusion: Intravenous gadoxetic acid can have a detrimental effect on im- Conclusion: CSI-MRI is an excellent approach to assess whole-body fat vol- age quality of triple arterial phases MR imaging using 4D eTHRIVE keyhole ume and organ fat content. It opens new perspectives for clinical research in technique. Third arterial phase images show best image quality regardless of imaging of metabolic disorders. the contrast media. S695 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 11:00 - 12:30 Wicklow Hall 2B SS 15.2 Diagnostic performance of dual-tracer positron Scientific Session SS 15 emission tomography-CT with 18F-fluorodeoxyglucose Liver and bile duct intervention: evaluating and 11C-acetate versus modified response evaluation criteria in solid tumors in the assessment of residual treatment response viable tumours in HCC that received loco-regional therapy SS 15.1 K.W.H. Chiu, T.-T. Cheung, R. Lo, H. Yuan, L.W. Chan; Preoperative CT findings for prediction of incomplete Hong Kong/HK resection and poor survival in patients with gallbladder Purpose: Dual-tracer positron emission tomography (DT-PET) with 11C-ace- cancer 1 2 1 1 1 tate (Ac) and 18F-fluorodeoxyglucose (FDG) have shown to be highly sensitive J.H. Kim , S.-Y. Choi , H.J. Park , H.W. Eun , J.K. Han ; 1 2 in the detection of HCC although thus far, it has only been applied to treat- Seoul/KR, Bucheon/KR ment-naïve patients. The aim of this study was to evaluate the diagnostic per- Purpose: To predict residual tumor (R) classification and poor survival in pa - formance DT-PET in assessing HCC that had undergone loco-regional treat- tients who underwent surgery for gallbladder cancer using preoperative CT. ment. Material and methods: From 2006 to 2012, 173 patients (M:F=96:77, Material and methods: Patients with HCC who received loco-regional treat- 65.3years) with gallbladder cancer underwent CT and surgery. Two radiolo- ment were retrospectively identified and included if they underwent CT or MRI gists assessed CT findings including tumor morphology, location, T-stage, ad - and DT-PET within 3 months of resection of the lesion. All lesions were evalu- jacent organ invasion, hepatic artery abutment, portal vein encasement, LN ated on cross-sectional imaging using the modified response evaluation crite - metastasis, metastasis, resectability, gallstone, and combined cholecystitis. R ria in solid tumours (mRECIST) and radiotracer uptake on PET were recorded classification was categorized into no residual tumor (R0) and residual tumor and compared with histology. (R1 or R2). We analyzed the correlation between CT findings and R classifica - Results: Fourteen patients with 30 HCC fulfilled the inclusion criteria; 11 un - tion. In addition, we followed up the patients up to 5 years and analyzed the derwent transarterial chemoembolization (TACE), 2 radiofrequency ablation relationship between each CT findings and overall survival using Cox regres - (RFA) and 1 both RFA and TACE (median time from previous intervention 3 sion analysis and Kaplan–Meier analysis. months, range 1-43 months). Residual tumour was present in 24 lesions on Results: R classification included no residual tumor in 127 patients and re - histology. Using mRECIST criteria, 16/30 lesions had complete response, 14 sidual tumor in 46 patients. 5-year survival was 44%. For prediction of R clas- partial response/stable disease. FDG was avid in 15/30, Ac in 20/30 and DT sification, hepatic invasion (Exp(B)=4.008, 95% CI, 1.22-6.39, p=0.014) and PET in 22/30 lesions. The positive predictive values for both modalities were hepatic artery abutment (Exp(B)= 8.67, 95% CI, 3.24-25.21, p=0.001) were 100% and the sensitivities in detecting residual disease were 54.2% and significant predictors for residual tumor. On survival analysis, residual tumor 83.3% (p<0.05) for cross-sectional imaging and DT-PET, respectively. (Exp(B)= 2.68, 95% CI, 1.72-4.19, p=0.001), liver metastasis (Exp(B)= 6.18, Conclusion: mRECIST criteria on cross-sectional imaging under-detect re- 95% CI, 1.41-27.07, p=0.016), duodenum invasion (Exp(B)= 2.8, 95% CI, 1.61- sidual viable tumour in HCC that had undergone loco-regional treatment. DT- 4.87, p=0.001), and regional LN metastasis (Exp(B)= 6.18, 95% CI, 1.41-27.07, PET is a sensitive modality that can be used as an alternative modality for the p=0.016) were significant predictors for poor survival. treatment reassessment especially in cases where patients’ management may Conclusion: Preoperative CT findings could aid planning surgery using the be altered. high-risk findings of residual tumor including hepatic invasion and hepatic ar - tery abutment. In addition, CT can predict poor survival using important find - SS 15.3 ings including liver metastasis, duodenum invasion, and regional LN metasta- CT texture analyses can predict early regrowth after sis. thermal ablation of colorectal liver metastases D.J. Van Der Reijd, E. Klompenhouwer, M. Taghavirazavizadeh, F. Staal, F. Imani, D. Meek, M.C. De Boer, D.M.J. Lambregts, R.G.H. Beets-Tan, M. Maas; Amsterdam/NL Purpose: To assess whether CT texture analysis can detect early regrowth of colorectal liver metastases (CRLM) one month after thermal ablation. Material and methods: 38 patients with CRLM who underwent radiofrequen- cy ablation (RFA) or microwave ablation (MWA) for lesions ≤3 cm were includ- ed. Ablation zones 1-month post-ablation were manually delineated on all slices on portal venous phase CT by an expert radiologist blinded to clinical outcome. A margin of 5 mm was mathematically added to the ablation zone and texture parameters (entropy, uniformity and mean grey-level intensity) were compared between patients with and without regrowth using multiple Laplacian of Gaussian filters ( σ0.5-2.5). Analyses were performed with pyradi- omics. Results: 18/38 patients had a regrowth. Mean follow-up was 32 (standard deviation (SD) 27) months and mean time to regrowth was 9 (SD 8) months. Entropy of the ablation zone (AZ) and ablation margin (AM) were significantly lower in regrowths (mean 1.34 vs 1.71, p=0.004 and mean 1.47 vs. 1.81, p=0.044, respectively). Uniformity of both the AZ and AM were higher in re- growths (mean 0.45 vs. 0.37, p=0.004 and mean 0.42 vs. 0.35, p=0.03, respec- tively). These differences were found both for analyses with a filter of σ0.5 and without a filter. For mean grey-level intensity no significant differences were observed for neither the AZ nor AM (mean 1.22 vs 0.73, p=0.496 and -1.90 vs -2.04, p=0.874, respectively). Conclusion: Patients with regrowth after ablation show a lower entropy and higher uniformity of the ablation zone and ablation margin 1-month post-abla- tion. CT texture analyses can help in identifying patients who are at risk for regrowth. S696 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 15.4 SS 15.7 Influence of pre-treatment tumor growth rate on the Factors associated with tumor progression after objective response of HCC treated with selective percutaneous ablation of HCC: results of a propensity trans-arterial chemoembolization score matching analysis comparing monopolar Y. Purcell, R. Pommier, V. Roche, M. Dioguardi Burgio, radiofrequency and microwaves A. Sibert, V. Vilgrain, M. Ronot; Clichy/FR D. Bouda, V. Barrau, C. Garcia-Alba, V. Roche, M. Dioguardi Burgio, A. Sibert, V. Vilgrain, M. Ronot; Purpose: To assess the influence of pre-treatment tumor growth rate ( TGR) pre Clichy/FR on the objective response (OR) according to modified response evaluation cri - teria in solid tumors (mRECIST) criteria after a first session of selective trans- Purpose: To identify risk factors for local and distant tumor progression after arterial chemoembolization (TACE) for the treatment of HCC. percutaneous thermal ablation of HCC with a focus on the comparison be- Material and methods: Ninety-nine patients (91 male, 93%) with 148 tumors tween microwave ablation (MWA) and monopolar radiofrequency ablation (mean 34±29mm), who underwent the first session of selective TACE for the (RFA). treatment of HCC between 2011 and 2016, were included. preTGR was calcu- Material and methods: Data were reviewed from consecutive cirrhotic pa- lated as the percentage change in tumour volume per month (%/mo) on avail- tients with very early/early HCCs who underwent image-guided percutaneous able CTs before treatment. OR was defined as a complete or partial response monopolar expandable RFA or MWA. Factors associated with local and distant on a lesion-based analysis. preTGR cut-off for prediction of OR was identified tumor progression were identified using propensity score matching (PSM). by receiver operating characteristic (ROC) curve analysis. Statistical analysis was performed with the Kaplan-Meier method using the Results: After one session of treatment 74 (50%), and 41 (28%) tumors log-rank test and Cox regression models. showed complete (CR), or partial response (PR) while 29 (20%) and 4 (2%) Results: A total of 190 HCC (mean 23±8.6mm) in 149 patients (mean 63±11 showed stable (SD), or progressive disease (PD). The mean TGR was 13±17 y-o, 79% male) underwent RFA (n=90, 47%) or MWA (n=100, 53%). Second- pre %/mo. It followed a U-shaped distribution with a mean 15±19, 13±16, 6±10, ary technical success rate was 97.4%. After a median follow-up of 24.6 and 15±5 %/mo in tumors showing CR, PR, SD, and PD. After exclusion of months (interquartile range (IQR): 9.7-37.2), 43 (23%) HCC showed local pro- tumors showing PD, TGR was significantly higher in tumors with OR gression (median time delay 13.4 months (IQR: 5.8-24.3)) and 91 (63%) pa- pre (mean14±18 %/mo) when compared to those showing SD (6±10 %/mo, tients had a distant progression (median time delay 10.4 months (IQR: 5.7-22). p=0.03). A TGR value of 6.5%/mo was associated with a sensitivity and Cox model after PSM identified treatment by RFA (hazard ratio (HR), 2.934; pre specificity of 60% and 86% for the prediction of OR (Area under ROC of P=0.006), size ≥30mm (HR, 3.130; P=0.007) and vascular contact (HR, 3.372; 0.67±0.05). P=0.005) as risk factors for local progression. Factors associated with distant Conclusion: Pre-treatment TGR is highly variable in HCC before TACE with a progression were a main tumor size ≥30mm (HR, 1.94; P=0.013), an alpha-fe- U-shaped distribution for the prediction of tumor response. It provides insight toprotein (AFP) level above 100 ng/mL (HR, 2.56; p=0.002), and hepatitis B into tumor biology that may be used during pre-treatment workup to help (HR, 0.51; p=0.047). stratify patients. Conclusion: The rate of local HCC progression was lower after MWA when compared to monopolar expandable RFA, regardless of tumor size and vascu- lar contact. Ablation technique did not influence the risk of distant progression. SS 15.5 withdrawn by the authors SS 15.8 Predicting the response of colorectal cancer liver SS 15.6 metastases to preoperative chemotherapy using Radiological tumor response and histopathological gadoxetic acid-enhanced MRI correlation of HCC treated with stereotactic body 1 2 2 2 N.V. Costa , N. Bastati , S. Pötter-Lang , Z. Guengoern , radiation therapy as a bridge to liver transplantation 2 2 1 2 Y. Bican , A. Ba-Ssalamah ; Lisbon/PT, Vienna/AT K. Mastrocostas, S. Fischer, P. Munoz-Schuffenegger, Purpose: The aim of the study was to create a scoring system based upon H.-J. Jang, G. Sapisochin, L.A. Dawson, T.K. Kim; ga gadoxetic acid-enhanced MRI ( MRI) features to predict the treatment re- Toronto, ON/CA sponse (TR) to chemotherapy. Purpose: To assess the imaging appearance of HCC treated with stereotactic Material and methods: This was a retrospective study of 30 consecutive pa- body radiation therapy (SBRT) as a bridging therapy prior to liver transplanta- tients (65.2±11.2years) with colorectal cancer liver metastases (CRCLM), who tion (LT), with pathological correlation. underwent gaMRI after chemotherapy and before hepatic resection. Metasta- Material and methods: 23 patients (20 males; mean age, 60 years) were in- ses were classified according to a suggested scoring system (0-6 points) in cluded, with a total of 25 lesions assessed. CT and/or MRI imaging pre- and three groups of response: optimal (≤2 points); partial (2-4 points); and no-re- post-SBRT were reviewed by two independent radiologists, with a third reader sponse (≥4 points). The scoring system comprised three features: overall ho- resolving inter-observer differences. The target lesions were assessed for ra- mogeneity (homogeneous=0, mixed=1, heterogeneous=2); tumor liver inter- diological response including change in size, change in attenuation in the por- face (sharp=0, mixed=1, ill-defined=2); and peripheral rim enhancement (≤2 tal venous phase on CT or signal intensity on MRI, and percentage necrosis if mm=0, 2-4 mm=1, ≥4 mm=2). Apparent diffusion coefficient (ADC) values were present. The radiological findings were compared to the tumor necrosis at measured. The primary outcome was residual vital tumor (RVT). The scoring explant as assessed by a pathologist. system, response groups, and ADC values were calculated and compared Results: Of the 25 lesions, 15 lesions were classed as stable disease (SD), 8 with the RVT percentage. Demographic, laboratory, and imaging findings were lesions as partial response (PR) and 2 lesions as progressive disease (PD). included in a multivariate statistical analysis. The three groups of response Among the 15 SD lesions, all lesions were identified as having necrosis at ex - were correlated with patient survival and the log-rank test was used to com- plant pathology and imaging. Ten of these lesions were assessed as having pare two survival distributions (optimal/partial responses vs no-response imaging changes of necrosis concordant within 10% of the necrosis at pathol- groups). ogy. Nine lesions were assessed at imaging as having >20% necrosis, all of Results: Forty-one CRCLM showed good inter-observer agreement (κ=0.86). which were concordant with the pathology findings, despite stable size. In only Multiple regression demonstrated an association between RVT (32.9±11.2) one case was necrosis at imaging >20% above that at explant pathology. and the scoring system (p<0.001), the response group (p<0.001), and the ADC Conclusion: Current standard HCC response criteria such as response evalu- values (p<0.021). The survival distributions between optimal/partial response ation criteria in solid tumors (RECIST) underestimate the response of lesions to and no-response showed a trend to be different (p=0.066). local therapy with SBRT. As the utility of SBRT as a focal therapy for HCC in- ga Conclusion: MRI correlated well with our scoring system, different response creases, further evaluation of the imaging findings of response is required. groups, and ADC values in patients with CRCLM treated with chemotherapy, and may be used to assess the RVT percentage. S697 Scientific Sessions Authors’ Index SCIENTIFIC SESSIONS / FRIDAY, JUNE 15, 2018 SS 15.9 SS 15.10 Early prediction of response to transarterial Early recurrence of HCC predicted by whole-lesion chemoembolization using CT perfusion in primary liver radiomics on preoperative gadoxetic acid-enhanced tumors MRI D. Tamandl, F. Waneck, A. Ba-Ssalamah, C. Loewe; Z. Zhang, J. Chen, H. Jiang, L. Cao, T. Duan, Z. Ye, B. Song; Vienna/AT Chengdu/CN Purpose: Assessment of CT perfusion (CTP) in the prediction of response Purpose: To investigate the role of gadoxetic acid (Gd-EOB-DTPA)-enhanced early after transarterial chemoembolization (TACE) in patients with HCC and MRI-derived texture features in predicting early recurrence (≤6 months) of cholangiocarcinoma (CCC). HCC. Material and methods: This prospective study was approved by the institu- Material and methods: 29 consecutive patients with preoperative Gd-EOB- tional review board and written informed consent was obtained from all pa- DTPA-enhanced MRI were included. Early recurrence was determined within 6 tients. We performed CTP one day before and after (interquartile range [IQR]: months. 3D texture features on individual sequence were derived using an in- 1-2 days) TACE in 18 patients. 46 lesions (41 HCC and 5 hypervascular CCC) house software (Analysis Kit, GE Healthcare). LASSO logistic regression mod- were amenable for analysis. Various perfusion-related parameters were meas- el was adopted for feature selection. Identified texture features were compared ured. In a lesion-based analysis, absolute and relative changes were com- between patients with and without early recurrence. Receiver-operating char- pared to the clinical response on biphasic CT using modified response evalu - acteristic (ROC) analysis of predicting early recurrence was performed for each ation criteria in solid tumors (mRECIST) criteria 6 weeks after the procedure. texture feature. Optimal cutoff values for detecting response were calculated using area under Results: 17 early recurrences were confirmed with imaging follow-up. The the ROC curves (AUROC). Follow-up analysis included assessment of overall skewness, sumEntropy, ShortRunHighGreyLevelEmphasis_AllDirection_off- survival (OS), local recurrence-free survival (LRFS) and local progression-free set7_SD, and LongRunLowGreyLevelEmphasis_AllDirection_offset7_SD of survival (LPFS), using Kaplan-Meier estimates. HCC on arterial-phase, the ShortRunHighGreyLevelEmphasis_AllDirection_ Results: The clinical response was: complete response (CR, n=22, 48%), par- offset4_SD and LongRunHighGreyLevelEmphasis_angle90_offset7 on portal tial response (PR, n=9, 20%) and no response (14 stable, 1 progression) 6 venous-phase (PVP), and uniformity and LongRunLowGreyLevelEmphasis_ weeks after TACE. CTP parameters were all reduced after TACE in responding AllDirection_offset7_SD on hepatobiliary phase (HBP) were found to be statis- patients (PR, CR) while no difference was observed in non-responders. Arte- tically different between patients with and without recurrence (P < 0.05). The rial liver perfusion (ALP ) was superior in prediction of CR compared to blood areas under the ROC curves (AUCs) were 0.966 (CI[0.82, 0.99]), 0.725 (CI[0.53, post flow (BF ) and blood volume (BV ) (AUROC 0.953 vs. AUROC 0.859 0.87]), 0.760 (CI[0.57, 0.89]) and 0.833 (CI[0.65, 0.95]) for the skewness, su- post post ALP BF and AUROC 0.831, p<0.001) with a sensitivity, specificity, positive predictive mEntropy, ShortRunHighGreyLevelEmphasis_AllDirection_offset7_SD, and BV value, negative predictive value and accuracy of 91%, 92%, 91%, 92% and LongRunLowGreyLevelEmphasis_AllDirection_offset7_SD on AP, 0.789 91%. Only 4/22 lesions with CR recurred with a median local recurrence-free (CI[0.59, 0.92]) and 0.814 (CI[0.63, 0.93]) for the ShortRunHighGreyLevelEm- survival of 22.7 months. phasis_AllDirection_offset4_SD and LongRunHighGreyLevelEmphasis_an- Conclusion: Early response assessment after transarterial chemoemboliza- gle90_offset7 on PVP, and 0.843 (CI[0.66, 0.99]) and 0.760 (CI[0.63, 0.93]) for tion can reliably be performed with CT perfusion. the uniformity and LongRunLowGreyLevelEmphasis_AllDirection_offset7_SD on HBP, respectively, in predicting early recurrence of HCC. The highest sensi- tivity and specificity were shown on arterial-phase images (skewness) of 91.7% (CI[61.5%,99.8%]) and 94.1% (CI[71.3%,99.9%]). Conclusion: Whole-lesion texture features based on Gd-EOB-DTPA-en- hanced-MRI held promise for predicting early recurrence of HCC after hepa- tectomy, especially the skewness on arterial-phase images. S698 Scientific Sessions Authors’ Index S699 Scientific Sessions Authors’ Index AUTHORS’ INDEX Beets-Tan R.G.H.: SS 2.4, SS 2.6, SS 4.3, Cao L.: SS 1.9, SS 9.3, SS 15.10 SS 4.10, SS 5.2, SS 9.1, SS 9.4, Caramia E.: SS 4.5 SS 9.9, SS 15.3 Cardano G.: SS 3.9, SS 8.7 Aalbers A.G.J.: SS 4.3 Behari J.: SS 7.2 Cardobi N.: SS 4.7, SS 8.4, SS 11.2 Aerts H.J.W.L.: SS 9.1, SS 9.4 Bektas C.T.: SS 5.9 Carrozzo V.: SS 5.1 Albazaz R.: SS 1.1, SS 13.3 Bellini D.: SS 5.8, SS 6.1, SS 12.10 Cartier V.: SS 7.1, SS 7.7 Albrecht M.H.: SS 4.8 Belmonte E.: SS 14.8 Caruso D.: SS 6.1 Aleman B.M.P.: SS 5.2 Bencardino D.: SS 14.5 Carvalho L.: SS 11.1 Alexander D.: SS 2.3 Benea G.: SS 12.9 Casiraghi A.: SS 6.10 Alfaro I.: SS 6.5 Berbee M.: SS 9.9 Cassinotto C.: SS 7.7 Amaddeo G.: SS 14.2 Bergamasco L.: SS 4.5, SS 5.1, SS 10.7 Castera L.: SS 7.5 Ambrosetti M.C.: SS 11.3 Bertuzzo L.: SS 3.9, SS 8.7, SS 8.10, Castro J.: SS 6.5 Andersson M.: SS 8.3 SS 11.8, SS 11.10 Catalano C.: SS 14.5 Anysz-Grodzicka A.: SS 9.5 Berzigotti A.: SS 13.8 Cazals-Hatem D.: SS 12.6 Aslan S.: SS 8.1 Beuzit L.: SS 7.3 Cecchin D.: SS 2.7 Atkinson D.: SS 6.2 Beykoz Çetin E.: SS 7.10 Chalaye J.: SS 14.2 Atri M.: SS 9.2 Bican Y.: SS 15.8 Chan L.W.: SS 15.2 Aubé C.: SS 7.1, SS 7.7, SS 13.7 Biscaldi E.: SS 12.4 Chang W.: SS 8.2 Auger M.: SS 7.3 Bissoli A.: SS 11.3 Chargeboeuf C.: SS 13.6 Axelsson E.: SS 8.3 Boddenberg J.: SS 11.9 Chen J.: SS 9.3, SS 10.4, SS 15.10 Ayuso C.: SS 1.5, SS 14.8 Bodelle B.: SS 4.8 Chermak F.: SS 7.7 Boellaard T.N.: SS 2.6 Cheung T.-T.: SS 15.2 Bogner P.: SS 6.7 Chiu K.W.H.: SS 15.2 Ba-Ssalamah A.: SS 5.4, SS 7.9, SS 13.5, Bonatti G.: SS 10.9 Cho S.B.: SS 14.7 SS 13.10, SS 15.8, SS 15.9 Bonatti M.: SS 10.9 Choi M.H.: SS 5.5 Badia S.: SS 5.8 Borg P.: SS 11.1 Choi S.-Y.: SS 1.8, SS 9.8, SS 15.1 Bae J.S.: SS 8.2, SS 9.8 Borhani A.: SS 7.2, SS 10.10 Chowdry P.: SS 5.10 Bakers F.C.H.: SS 2.4, SS 4.10, SS 9.1, Bouda D.: SS 15.7 Christe A.: SS 13.8 SS 9.4, SS 9.9 Boudiaf M.: SS 6.6 Chryssou E.: SS 1.10 Baldwin-Cleland R.: SS 12.1 Boulay-Coletta I.: SS 3.4, SS 3.8 Churilov L.: SS 5.10 Bali M.A.: SS 5.8 Bouquot M.: SS 6.6 Ciaravino V.: SS 4.7, SS 11.2 Ball C.: SS 12.7 Boursier J.: SS 7.1, SS 7.7, SS 13.7 Cieszanowski A.: SS 9.5 Baltzer P.: SS 13.10 Bowden D.: SS 3.7 Cingarlini S.: SS 11.2 Bannier E.: SS 13.7 Bowden D.J.: SS 3.2 Clarke C.G.D.: SS 1.1, SS 12.2 Baranes L.: SS 14.2, SS 14.4 Bozzato A.M.: SS 4.5 Çolakoğlu M.K.: SS 7.10 Barat M.: SS 3.5, SS 6.6 Bradley K.: SS 8.6 Collins C.: SS 5.3 Barata M.J.: SS 2.5, SS 2.8 Brown P.J.: SS 2.1, SS 4.9 Conci S.: SS 4.7 Barbieri S.: SS 12.9 Bruix J.: SS 1.5, SS 14.8 Contro A.: SS 4.7 Bardou-Jacquet E.: SS 13.6, SS 13.7 Brun V.: SS 7.3 Conwell D.: SS 11.7 Barra F.: SS 12.4 Brunetti F.: SS 14.4 Corno L.: SS 3.4, SS 3.8 Barrau V.: SS 15.7 Bruno O.: SS 1.7 Corr A.: SS 2.9 Bartels-Rutten A.: SS 5.2 Brusic A.: SS 5.10 Costa F.: SS 11.1 Bartholomä W.: SS 8.3 Bruzzi J.: SS 5.3 Costa N.V.: SS 15.8 Bassi C.: SS 8.9 Buonocore V.: SS 14.5 Cournane S.: SS 4.2 Bassi M.: SS 12.9 Burling D.: SS 2.9, SS 12.1 Cressoni M.: SS 4.1 Bastati N.: SS 15.8 Byrne D.: SS 6.8 Crimì F.: SS 2.7 Bastati-Huber N.: SS 7.9, SS 13.5, Byrne J.: SS 3.3 Crona J.: SS 11.1 SS 13.10 Cronin C.: SS 6.8 Bazeries P.: SS 7.1 Cronin C.G.: SS 4.4 Beable R.: SS 12.7 Cui J.: SS 9.10 Calame P.: SS 3.5 Beer A.: SS 7.9 Cybulski A.J.: SS 3.9 Cales P.: SS 7.7 Beer L.B.: SS 13.5 Camlidag İ.: SS 8.1 Beets G.L.: SS 2.6, SS 9.4 Cannella R.: SS 7.2, SS 10.10 Cantwell C.P.: SS 3.10 S700 Scientific Sessions Authors’ Index AUTHORS’ INDEX Grimm R.: SS 8.4 D F Grossarth V.: SS 11.9 Guengoern Z.: SS 15.8 D’Onofrio M.: SS 4.7, SS 8.4, SS 8.9, Fabiszewska E.: SS 9.5 Guimaraes L.: SS 14.6 SS 11.2 Fabris C.: SS 8.10, SS 11.3, SS 11.10 Guler E.: SS 3.1 Darnell A.: SS 1.5, SS 14.8 Faccinetto A.: SS 7.5 Gursoy Coruh A.: SS 2.10 Darwish O.: SS 7.6 Fahlenkamp U.: SS 10.5 Guthrie J.A.: SS 1.1, SS 13.3 Dasyam A.: SS 11.7 Faki A.A.: SS 3.6 Dautry R.: SS 6.6 Faletti R.: SS 4.5, SS 5.1, SS 10.7 David M.: SS 12.6 Faluhelyi N.: SS 6.7 Dawson L.A.: SS 15.6 Famularo S.: SS 13.4 Haimerl M.: SS 13.2 De Angelis N.: SS 14.4 Farkas O.: SS 6.7 Halligan S.: SS 12.1 De Arcos J.: SS 7.6 Feeney J.: SS 12.3 Hamilton G.: SS 9.10 De Boer M.C.: SS 15.3 Fehrenbach U.: SS 10.5 Hammerstingl R.: SS 4.8 De Ledinghen V.: SS 7.7 Feier D.S.: SS 7.9 Han J.K.: SS 8.2, SS 9.8, SS 15.1 De Mestier L.: SS 11.1 Fenlon H.: SS 4.4, SS 6.8 Hanbidge A.: SS 3.3 De Paoli Barbato G.: SS 12.9 Ferraris A.: SS 4.5 Harguem S.: SS 10.8 De Robertis R.: SS 4.7, SS 8.4, SS 8.9, Ferrero S.: SS 12.4 Harman M.: SS 3.1 SS 11.2 Fidler J.L.: SS 8.6 Hatzidakis A.A.: SS 1.10, SS 4.6 De Santis D.: SS 6.1 Fighera A.: SS 3.9 Haug A.: SS 5.4 Del Chiaro M.: SS 8.3 Figueiredo N.: SS 2.5, SS 2.8 Healy G.M.: SS 3.10 Delahaye J.: SS 7.1 Fiore S.: SS 4.5 Heidous M.: SS 3.6 Deledda A.: SS 12.9 Fischer S.: SS 15.6 Henderson W.C.: SS 9.10 Delli Pizzi A.: SS 2.6 Fletcher J.G.: SS 8.6 Hérin E.: SS 14.2, SS 14.4 Denecke T.: SS 10.5 Fogel E.L.: SS 11.5, SS 11.7 Heverhagen J.: SS 13.8 Dijkhoff R.A.P.: SS 2.4, SS 4.10 Fonio P.: SS 4.5, SS 5.1, SS 10.7 Higginson A.: SS 12.7 Dioguardi Burgio M.: SS 7.5, SS 15.4, Forner A.: SS 1.5, SS 14.8 Hill A.: SS 3.2, SS 3.7 SS 15.7 Fouchard I.: SS 7.7 Hiriart J.-B.: SS 7.7 Djabbari M.: SS 14.4 Franca R.: SS 11.1 Hoeffel C.: SS 6.6 Doblas S.: SS 12.6 Frantsev D.: SS 10.6 Hoffmann R.-T.: SS 11.6, SS 14.10 Dohan A.: SS 6.6 Friedrich N.: SS 11.6, SS 14.10 Hooker J.C.: SS 9.10 Dolgushin B.: SS 10.6 Frigerio I.: SS 8.9 Houssel-Debry P.: SS 7.3 Drago S.G.: SS 2.4, SS 4.1, SS 4.10, Fritz F.: SS 8.5 Huber A.T.: SS 13.8 SS 6.10 Frostberg E.: SS 12.8 Hupkens B.J.P.: SS 2.6 Dromain C.: SS 11.1 Fukuda H.: SS 10.3 Hwang J.A.: SS 1.8 Duan T.: SS 1.9, SS 10.4, SS 15.10 Furlan A.: SS 7.2, SS 10.10 Dubois M.: SS 7.3 Füger B.: SS 5.4 Dunne R.: SS 3.2, SS 3.7 Đuri ć-Stefanović A.: SS 8.8 Ianus A.: SS 2.2, SS 2.3 Ibukuro K.: SS 10.3 Gandon Y.: SS 7.3, SS 13.6, SS 13.7 Imani F.: SS 15.3 Garcia-Alba C.: SS 15.7 Imbault M.: SS 7.5 Effler J.: SS 14.10 García-Criado A.: SS 1.5 Iori A.P.: SS 14.5 Einspieler H.: SS 7.9 Garg I.: SS 8.6 Ippolito D.: SS 4.1, SS 6.10, SS 13.4 Elkrief L.: SS 13.9 Garteiser P.: SS 12.6 Itti E.: SS 14.2 Elmas N.Z.: SS 3.1 Gatti M.: SS 4.5, SS 5.1, SS 10.7 Elmer M.: SS 13.5 Geisel D.: SS 10.5 Emsen B.: SS 14.2 Gennisson J.-L.: SS 7.5 Engbersen M.: SS 4.3 Gianotti L.: SS 13.4 Jain R.: SS 7.4 Erden A.: SS 2.10 Gibney R.: SS 1.3, SS 10.2 Jang H.-J.: SS 9.2, SS 13.1, SS 15.6 Ernst O.: SS 3.8 Giganti M.: SS 12.9 Jasieniak J.: SS 9.5 Esler S.: SS 5.10 Girelli R.: SS 8.9 Javed H.: SS 14.6 Eun H.W.: SS 15.1 Goenka A.: SS 8.6 Jenkins J.T.: SS 2.9 Eveno C.: SS 6.6 Gołębiewski B.: SS 9.5 Jeon C.Y.: SS 11.5 Gollifer R.: SS 6.2 Jeon S.J.: SS 9.8 Greer P.: SS 11.7 Jeong W.K.: SS 1.4, SS 10.1 S701 Scientific Sessions Authors’ Index AUTHORS’ INDEX Jevtić S.: SS 8.8 Laghi A.: SS 5.8, SS 6.1, SS 12.10 Mansueto G.: SS 3.9, SS 8.7, SS 8.10, SS 11.3, SS 11.8, SS 11.10 Jiang H.: SS 9.3, SS 15.10 Lahaye M.J.: SS 4.3, SS 9.1, SS 9.4, SS 9.9 Marchegiani G.: SS 3.9, SS 8.10 Johnson G.B.: SS 8.6 Johnson S.: SS 9.2 Lamarca A.: SS 11.1 Maretto I.: SS 2.7 Lambregts D.M.J.: SS 2.4, SS 2.6, Martel A.: SS 9.2 Johnston C.: SS 4.2 SS 4.3, SS 4.10, SS 5.2, SS 9.1, Joo I.: SS 8.2 Martin S.S.: SS 4.8 SS 9.4, SS 9.9, SS 15.3 Marx C.: SS 13.8 Jouan J.: SS 13.6, SS 13.7 Landoni L.: SS 11.2 Jung S.E.: SS 5.5 Mastrocostas K.: SS 15.6 Laniado M.: SS 11.6, SS 14.10 Masulovic D.: SS 8.8 K Lannes A.: SS 7.7 Matos C.: SS 2.2, SS 2.3, SS 2.5, SS 2.8, Lapteva M.: SS 10.6 SS 9.7 Kaci R.: SS 6.6 Lapuyade B.: SS 7.7 Mayer P.F.J.: SS 8.5 Kadıoğlu M.E.: SS 7.10 Laqua R.: SS 14.10 McCann J.W.: SS 3.10 Kalarakis G.: SS 1.10, SS 4.6 Lauenstein T.: SS 11.9 McCormick P.A.: SS 1.3, SS 10.2 Kalcan S.: SS 7.10 Laurent A.: SS 14.2, SS 14.4 McLoughlin L.: SS 4.2 Kale S.: SS 7.4 Lawlor R.: SS 9.2 McQuade C.S.: SS 4.4 Karantanas A.: SS 1.10, SS 4.6 Lazić L.: SS 8.8 Meek D.: SS 15.3 Kartalis N.: SS 8.3 Le Bail B.: SS 7.7 Melisi D.: SS 8.9 Kauczor H.U.: SS 8.5 Lebert P.: SS 3.8 Menys A.: SS 6.2 Kavanagh R.G.: SS 5.6 Lebigot J.: SS 7.1, SS 7.7 Mertineit N.: SS 13.8 Kazemi-Shirazi L.: SS 13.10 Lee A.: SS 3.2, SS 3.7 Metin Y.: SS 7.10 Keane R.: SS 1.3 Lee D.H.: SS 5.7, SS 7.8 Metric Investigators T.: SS 6.9 Keussen I.: SS 8.3 Lee J.Y.: SS 7.8 Michalak S.: SS 7.7 Khalili K.: SS 9.2 Lee M.: SS 3.2, SS 3.7, SS 12.5 Middleton M.S.: SS 9.10 Kiani A.: SS 3.5 Lee S.: SS 10.1 Miles A.: SS 6.9 Kilickesmez O.: SS 5.9 Lee S.M.: SS 5.7 Millet I.: SS 3.4, SS 3.8 Kim J.E.: SS 6.4, SS 9.8 Leithner D.: SS 4.8 Milliat F.: SS 12.6 Kim J.H.: SS 8.2, SS 9.8, SS 15.1 Lenga L.: SS 4.8 Min J.H.: SS 1.4 Kim K.W.: SS 10.1 Li L.: SS 11.5 Minervini M.I.: SS 10.10 Kim M.J.: SS 14.7 Li M.: SS 11.4 Moghe S.: SS 14.6 Kim S.H.: SS 5.7, SS 6.4 Liddy S.: SS 12.3 Moloney F.: SS 5.6 Kim S.S.: SS 1.8 Lin X.: SS 10.4 Mondal D.: SS 14.6 Kim T.K.: SS 9.2, SS 13.1, SS 15.6 Lo R.: SS 15.2 Monnet A.: SS 14.2 Kim T.O.: SS 6.4 Loewe C.: SS 15.9 Montesano M.: SS 12.10 Kim Y.K.: SS 1.4 Loizou L.: SS 8.3 Monti M.L.: SS 11.8 Kishino M.: SS 10.3 Lombardi S.: SS 6.10 Morrin M.: SS 3.2, SS 3.7, SS 12.5 Klauss M.: SS 8.5 Lombardo F.: SS 10.9 Mouries A.: SS 7.7 Klompenhouwer E.: SS 15.3 Lopez C.: SS 11.1 Muin D.: SS 13.10 Kocak B.: SS 5.9 Lopez M.: SS 14.5 Mulé S.: SS 14.2, SS 14.4 Kok N.F.M.: SS 4.3 Lourenço J.M.G.: SS 9.7 Mulsow J.: SS 4.4 Kokkinos N.: SS 4.6 Luciani A.: SS 14.2, SS 14.4 Munoz-Schuffenegger P.: SS 15.6 Kose T.: SS 3.1 Lundell L.: SS 8.3 Murphy A.N.: SS 4.4 Kosidekakis N.: SS 4.6 Murray T.É.: SS 12.5 Kovac J.: SS 8.8 Kromrey M.-L.: SS 11.6, SS 14.10 M B.: SS 7.4 Kühn J.-P.: SS 11.6, SS 14.10 Maas M.: SS 2.4, SS 2.6, SS 4.10, Na Yeon H.: SS 14.7 Kukołowicz P.: SS 9.5 SS 5.2, SS 9.1, SS 9.4, SS 9.9, Najran P.S.: SS 11.1 Kumar D.: SS 3.6 SS 15.3 Nederveen A.J.: SS 7.6 Kundaragi N.: SS 7.4 Maccioni F.: SS 14.5 Negrelli R.: SS 8.10 Kuś P.: SS 9.5 Maher M.: SS 5.6 Nikolić T.: SS 8.8 Maher P.: SS 5.10 L Nolan N.: SS 1.3, SS 10.2 Maheux A.: SS 10.8 Nural M.S.: SS 8.1 Lacognata C.: SS 2.7 Malone D.E.: SS 1.3, SS 10.2 Lacroix M.: SS 14.4 Manhoopi Y.: SS 12.8 Lagadec M.: SS 3.5 S702 Scientific Sessions Authors’ Index AUTHORS’ INDEX Plumb A.: SS 6.2, SS 12.1 O S Pocard M.: SS 6.6 Podgorska J.: SS 9.5 O Brien C.M.: SS 10.2 Saint-Jalmes H.: SS 13.6 Poelsterl S.: SS 13.2 O’Brien A.C.: SS 1.3, SS 3.10, SS 10.2 Salvia R.: SS 8.9 Polcaro A.: SS 7.6 O’Brien C.: SS 1.3 Sammon J.: SS 3.3, SS 14.6 Pomerri F.: SS 2.7 O’Brien S.: SS 5.6 Samreen N.: SS 8.6 Pommier R.: SS 15.4 O’Connor O.J.: SS 5.6 Sankar M.: SS 11.5 Pötter-Lang S.: SS 7.9, SS 13.5, O’Dwyer E.: SS 12.3 Santiago I.: SS 2.2, SS 2.3, SS 2.5, SS 13.10, SS 15.8 O’Leary P.: SS 3.7 SS 2.8, SS 9.7 Pourafkari M.: SS 9.2 Santinha J.: SS 2.2, SS 4.10, SS 9.7 O’Malley E.: SS 5.3 Pozzi Mucelli R.: SS 8.7, SS 11.8, Sapena V.: SS 14.8 O’Neill A.: SS 12.3 SS 11.10 O’Rourke C.: SS 4.2 Sapisochin G.: SS 15.6 Prasad V.: SS 10.5 Sarkar R.: SS 12.3 O’Shea A.: SS 12.5 Purcell Y.: SS 15.4 O’Suilleabhain C.: SS 5.6 Sarno A.: SS 4.7, SS 8.9, SS 11.2 Sartoris R.: SS 13.9 Obaro A.E.: SS 12.1 Oberti F.: SS 7.7 Savoldi E.: SS 12.7 Scala C.: SS 12.4 Obmann V.C.: SS 13.8 Quehen E.: SS 7.3 Scarpa A.: SS 11.2 Ogasawara G.: SS 10.3 Ogier-Denis E.: SS 12.6 Schaefer N.: SS 11.1 Schneider T.: SS 7.6 Okrainec A.: SS 3.3 Rafaelsen S.R.: SS 12.8 Opalinska M.: SS 11.1 Schurink N.: SS 9.9 Rahmouni A.: SS 14.2, SS 14.4 Segersvärd R.: SS 8.3 Ordas I.: SS 6.3 Orhan Metin N.: SS 7.10 Rahr H.: SS 12.8 Seltman T.A.: SS 11.5 Raimondi E.: SS 12.9 Sergeeva O.: SS 10.6 Orsini E.B.: SS 13.4 Ramanathan S.: SS 3.6 Shah Z.: SS 11.7 Özdemir O.: SS 7.10 Ramotar H.: SS 13.3 Ozutemiz O.: SS 3.1 Sheehan M.: SS 3.2, SS 3.7 Rangaswamy B.: SS 10.10 Sheehy N.: SS 4.2 P Rautou P.-E.: SS 1.7, SS 7.5, SS 13.9 Shemesh N.: SS 2.2, SS 2.3 Rayar M.: SS 7.3 Sheridan M.B.: SS 1.1, SS 13.3 Paiella S.: SS 8.9 Reddy N.: SS 7.4 Shields C.: SS 4.4 Paisant A.: SS 7.3, SS 13.6, SS 13.7 Regnault H.: SS 14.2 Shili S.: SS 7.7 Pałucki J.: SS 9.5 Rengo M.: SS 5.8, SS 12.10 Shin H.C.: SS 1.8 Panchal N.: SS 7.4 Reynolds J.V.: SS 4.2 Shorikov M.A.: SS 10.6 Panés J.: SS 6.3, SS 6.5 Ricart E.: SS 6.3, SS 6.5 Shur J.: SS 2.9 Papadakis A.: SS 1.10 Richard P.: SS 14.4 Sibert A.: SS 15.4, SS 15.7 Papanikolaou N.: SS 2.2, SS 4.10, SS 9.7 Rimola J.: SS 1.5, SS 6.3, SS 6.5, Sim K.C.: SS 14.7 Parés O.: SS 2.5, SS 2.8 SS 14.8 Sinkus R.: SS 7.6 Park B.J.: SS 14.7 Rivosecchi F.: SS 6.1 Sirlin C.: SS 9.10 Park H.J.: SS 15.1 Rizzati R.: SS 12.9 Sironi S.: SS 4.1, SS 6.10 Park S.J.H.: SS 9.8 Roche V.: SS 15.4, SS 15.7 Skehan S.J.: SS 1.3, SS 10.2 Pasicz K.: SS 9.5 Rodriguez S.: SS 6.3 Skornitzke S.: SS 8.5 Patil A.: SS 3.10 Roe C.: SS 12.2 Skrzyński W.: SS 9.5 Pautrat K.: SS 6.6 Rogalla P.: SS 14.6 Sleeman M.: SS 5.10 Pavel M.: SS 10.5, SS 11.1 Rondenet C.: SS 3.4 Smith C.R.: SS 1.1, SS 13.3 Pecorelli A.: SS 4.1, SS 13.4 Ronot M.: SS 1.7, SS 3.5, SS 7.5, Song B.: SS 1.9, SS 9.3, SS 10.4, Pedersen M.R.: SS 12.8 SS 10.8, SS 11.1, SS 13.9, SS 15.4, SS 15.10 Peker E.: SS 2.10 SS 15.7 Soyer P.: SS 6.6 Perin A.: SS 2.7 Rossington H.: SS 2.1, SS 4.9 Staal F.: SS 15.3 Perisinakis K.: SS 1.10, SS 4.6 Rousset S.: SS 10.7 Staufer K.: SS 13.10 Persohn S.A.: SS 11.5 Rowe I.: SS 1.1 Stemmer A.: SS 8.4 Pezzutti D.: SS 11.1 Runge J.H.: SS 7.6 Stiller W.: SS 8.5 Picchia S.: SS 5.8, SS 12.10 Rutledge N.: SS 3.10 Stirling A.D.: SS 4.4 Pierce B.: SS 1.3 Ruzzenente A.: SS 4.7 Stoker J.: SS 6.2 Pigneur F.: SS 14.2, SS 14.4 Ryan R.: SS 1.3, SS 10.2 Stoyanova D.P.: SS 13.5 Plessier A.: SS 1.7 Strinnholm J.: SS 8.3 S703 Scientific Sessions Authors’ Index AUTHORS’ INDEX Stroszczynski C.: SS 13.2 Van’t Sant I.: SS 4.3 Zerunian M.: SS 6.1 Sundin A.: SS 11.1 Vas D.: SS 6.5 Zhang Z.: SS 15.10 Sung D.J.: SS 14.7 Vasin D.V.: SS 8.8 Ziayee N.: SS 11.9 Szeverenyi N.: SS 9.10 Vellone V.: SS 12.4 Zins M.: SS 3.4, SS 3.8 Verfürth F.: SS 11.9 Zucchetta P.: SS 2.7 Vermersch M.: SS 14.2 Vidal Trueba H.: SS 11.1 Taghavirazavizadeh M.: SS 15.3 Vietti Violi N.: SS 11.1 Takahashi N.: SS 11.7 Vilana R.: SS 1.5 Talei Franzesi C.: SS 4.1, SS 6.10, Vilgrain V.: SS 1.7, SS 3.5, SS 7.5, SS 13.4 SS 10.8, SS 12.6, SS 13.9, SS 15.4, Talwade R.: SS 7.4 SS 15.7 Tamandl D.: SS 5.4, SS 13.5, SS 13.10, Vipperla K.: SS 11.7 SS 15.9 Virshke E.: SS 10.6 Tang G.: SS 11.7 Vitale D.: SS 14.5 Tanter M.: SS 7.5 Vliegen R.F.A.: SS 9.1 Taourel P.: SS 3.4, SS 3.8 Vogl T.J.: SS 4.8 Taylor S.A.: SS 6.2, SS 6.9 Vollenbrock S.E.: SS 5.2 Tedesco G.: SS 4.7, SS 8.9, SS 11.2 Voncken F.E.M.: SS 5.2 Ter Beek L.C.: SS 5.2 Vos F.: SS 6.2 Territo P.R.: SS 11.5 Thornton E.: SS 12.5 Tilli M.: SS 12.9 Tinazzi Martini P.: SS 8.4, SS 11.2 Wald R.M.: SS 13.1 Tirkes T.: SS 11.5, SS 11.7 Walker S.P.: SS 13.3 Tobe K.: SS 10.3 Walsh J.P.: SS 6.8 Tolan D.J.M.: SS 2.1, SS 4.9, SS 12.2 Waneck F.: SS 15.9 Topazian M.: SS 11.7 Wang M.: SS 1.9 Treanor D.: SS 1.1 Waters C.: SS 5.3 Trebeschi S.: SS 9.4 Welaratne I.: SS 4.2 Troelstra M.A.: SS 7.6 Wichmann J.L.: SS 4.8 Truty M.: SS 8.6 Wieszczy P.: SS 9.5 Tsung A.: SS 10.10 Wiggermann P.: SS 13.2 Tublin M.: SS 7.2 Wilson D.: SS 13.3 Turkcanoglu M.H.: SS 5.9 Wyatt J.I.: SS 1.1 Turlin B.: SS 7.3 Twomey M.: SS 5.6, SS 14.6 Yadav D.: SS 11.7 Yang H.K.: SS 13.1 Ugarte-Cano C.: SS 12.1 Yang N.: SS 5.10 Yardımcı A.H.: SS 5.9 Ye Z.: SS 1.6, SS 15.10 Yoo S.-J.: SS 13.1 Valletta R.: SS 10.9 Yoon J.-H.: SS 6.4 Valls C.: SS 8.3 Yu H.: SS 11.4 Van Beers B.: SS 7.5, SS 12.6 Yuan H.: SS 15.2 Van Der Reijd D.J.: SS 15.3 Van Der Sande M.: SS 2.6 Van Dieren J.M.: SS 5.2 Van Eden H.: SS 4.3 Zagórowicz E.: SS 9.5 Van Elmpt W.J.C.: SS 9.9 Zamboni G.A.: SS 3.9, SS 8.7, SS 8.10, Van Griethuysen J.: SS 2.4, SS 2.6, SS 10.9, SS 11.3, SS 11.8, SS 11.10 SS 4.10, SS 9.9 Zanirato M.: SS 8.4 Van Griethuysen J.J.M.: SS 9.1, SS 9.4 Zappa M.: SS 3.5, SS 12.6 Van Wettere M.: SS 1.7 Zegai B.: SS 14.4 S704

Journal

Insights into ImagingSpringer Journals

Published: May 29, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off