Objectives To compare the image quality of the lungs between ultra-high-resolution CT (U-HRCT) and conventional area detector CT (AD-CT) images. Methods Image data of slit phantoms (0.35, 0.30, and 0.15 mm) and 11 cadaveric human lungs were acquired by both U-HRCTand AD-CT devices. U-HRCT images were obtained with three acquisition modes:normalmode(U-HRCT : 896 channels, 0.5 mm × 80 rows; 512 matrix), super-high-resolution mode (U-HRCT : 1792 channels, 0.25 mm × 160 rows; 1024 matrix), and volume SHR mode (U-HRCT : non-helical acquisition with U-HRCT ). AD-CT images were obtained with the same conditions as SHR-VOL SHR U-HRCT . Three independent observers scored normal anatomical structures (vessels and bronchi), abnormal CT findings (faint nodules, solid nodules, ground-glass opacity, consolidation, emphysema, interlobular septal thickening, intralobular reticular opac- ities, bronchovascular bundle thickening, bronchiectasis, and honeycombing), noise, artifacts, and overall image quality on a 3-point scale(1=worst, 2=equal, 3=best)comparedwith U-HRCT . Noise values were calculated quantitatively. Results U-HRCT could depict a 0.15-mm slit. Both U-HRCT and U-HRCT significantly improved visualization of SHR SHR-VOL normal anatomical structures and abnormal CT findings, except for intralobular reticular opacities and reduced artifacts, com- pared with AD-CT (p < 0.014). Visually, U-HRCT has less noise than U-HRCT and AD-CT (p < 0.00001). SHR-VOL SHR Quantitative noise values were significantly higher in the following order: U-HRCT (mean, 30.41), U-HRCT SHR SHR-VOL (26.84), AD-CT (16.03), and U-HRCT (15.14) (p < 0.0001). U-HRCT and U-HRCT resulted in significantly higher N SHR SHR-VOL overall image quality than AD-CT and were almost equal to U-HRCT (p <0.0001). Conclusions Both U-HRCT and U-HRCT can provide higher image quality than AD-CT, while U-HRCT was SHR SHR-VOL SHR-VOL less noisy than U-HRCT . SHR Key Points � Ultra-high-resolution CT (U-HRCT) can improve spatial resolution. � U-HRCT can reduce streak and dark band artifacts. � U-HRCT can provide higher image quality than conventional area detector CT. � In U-HRCT, the volume mode is less noisy than the super-high-resolution mode. � U-HRCT may provide more detailed information about the lung anatomy and pathology. . . . . Keywords Multidetector computed tomography Diagnostic imaging Lung diseases Image enhancement Artifacts Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00330-018-5491-2) contains supplementary Abbreviations material, which is available to authorized users. AD-CT Conventional area detector computed tomography * Masahiro Yanagawa ADIR 3D Adaptive iterative dose reduction m–firstname.lastname@example.org–u.ac.jp in three dimensions CT Computed tomography Department of Radiology, Osaka University Graduate School of CTDIvol Volumetric computed Medicine, 2-2 Yamadaoka, Suita-city, Osaka 565-0871, Japan tomography dose index Department of CT Systems, Canon Medical Systems Corp., HU Hounsfield units Otawara, Tochigi, Japan Eur Radiol (2018) 28:5060–5068 5061 MDCT Multidetector row radiation dose and focal spot size, a smaller focus size has computed tomography become operational on the U-HRCT device. U-HRCT Ultra-high-resolution The most advantageous feature of U-HRCT is its improved computed tomography spatial resolution (120 micron) , which makes finer features U-HRCT Ultra-high-resolution computed distinguishable on CT images. No study so far has evaluated tomography with normal mode the image quality of current U-HRCT acquisitions. The pur- U-HRCT Ultra-high-resolution computed pose of this study was to compare the image quality of the SHR-VOL tomography with volume mode lungs between U-HRCT and conventional AD-CT. U-HRCT Ultra-high-resolution computed SHR tomography with super-high- resolution mode Materials and methods This study was approved by the internal Ethics Review Board Introduction of our institute. Informed consent for the retrospective review of patient records and images and use of patient biomaterial From the introduction of the first CT device in 1972 to the was waived. present day , computed tomography (CT) has become an essential imaging modality in a wide range of clinical Phantom study applications through the incorporation of several innova- tive technologies. Thin-slice images of the whole lungs can Image data of slit phantoms (Kyoto Kagaku Corp., Kyoto, be easily obtained within one breath hold by multidetector- Japan) made of stainless steel were acquired to evaluate the row computed tomography (MDCT) . Both hardware spatial resolution of AD-CT (Aquilion ONE™;Canon and software have been developed to acquire image data Medical Systems Corp., formerly Toshiba Medical Systems, of wide spatial ranges in a short time, facilitated by faster Otawara, Tochigi, Japan) and U-HRCT (Aquilion gantry rotation speeds, a widening of detectors [e.g., 320- Precision™; Canon Medical Systems Corp., Otawara, detector-row CT systems with area detectors (AD-CT)], Tochigi, Japan). The slits (0.35, 0.30, and 0.15 mm) and in- higher generator power and increased stability of X-ray tervening spaces were the same width in each phantom. tubes, and detectors (e.g., garnet-based detectors) available Image data of each phantom were acquired with both U- in clinical settings [3–6]. Although the evolution toward HRCT and AD-CT. The common acquisition parameters were faster scanning of a wider range is remarkable, there has as follows: gantry rotation period, 1.5 s; X-ray voltage, 120 been little progress in increasing spatial resolution over the kV ; tube current, 200 mA; field of view, 20 mm. The protocol last 30 years. for AD-CT was as follows: the number of channels per detec- Regarding the spatial resolution, Imai et al. previously dem- tor row, 896 channels and 0.5 mm × 4 rows; matrix size, 512. onstrated improvements in spatial resolution using an experi- The U-HRCT was used in super-high-resolution mode (U- mental MDCT equipped with a density-double matrix detector HRCT : the number of channels per detector row, 1792 SHR [1824 channels (x-y plane) × 32 rows (z-axis) at a row width of channels and 0.25 mm × 4 rows; matrix size, 1024). Axial 0.3125 mm] and an X-ray tube with an ultra-small focal spot thin-section CT images of 0.5 mm thickness were reconstruct- . This experimental CT device provided high-resolution im- ed using a lung kernel (FC81): the frequency range of the lung aging while maintaining low-contrast detectability, suggesting a kernel for U-HRCT was twice as wide as that for AD-CT potential for clinical use in areas requiring high spatial resolu- because the number of channels of the U-HRCT device was tion, such as imaging of the inner ear, lungs, and bone. In 2017, twice that of the AD-CT device. the ultra-high-resolution CT (U-HRCT) device became avail- able for clinical practice. Kakinuma et al. reportedonthe Cadaveric human lungs and imaging performance of a U-HRCT prototype: a 4-row CT device with a detector element size of 0.25 × 0.25 mm at the isocenter and a Eleven cadaveric human lungs were inflated and fixed using the beam collimation of 0.25 mm × 4 rows. The detector element Heitzman method . These lungs were distended through the size of the U-HRCT is half that of a conventional AD-CT in main bronchus with fixative fluid that contained polyethylene both the in-plane and body-axis directions. Current U-HRCT glycol 400, 95% ethyl alcohol, 40% formalin, and water in devices have detectors with 1792 channels in 160 rows. The proportions of 10:5:2:3. The specimens were immersed in fix- minimum focus size (0.4 × 0.5 mm) is about a third of the area ative fluid for 2 days and then air-dried. The pathological diag- of a conventional ADCT device (0.9 × 0.8 mm), and the X-ray noses of these 11 lungs were: pulmonary hemorrhage (n =1), tube has also improved compared with conventional ADCT cardiogenic edema (n = 1), diffuse panbronchiolitis (n =1), devices. By optimizing the relationship between the required pulmonary tuberculosis (n = 2), pulmonary emphysema (n = 5062 Eur Radiol (2018) 28:5060–5068 1), diffuse alveolar damage (n = 1), pulmonary metastasis (n = (HU) and a window width of 1600 HU. The radiologists inde- 1), pulmonary lymphangitic carcinomatosis (n =1), andusual pendently evaluated abnormal CT findings (faint nodules, solid interstitial pneumonia (n =2). nodules, ground-glass opacity, consolidation, emphysema, in- Image data of the 11 lungs were acquired with both U- terlobular septal thickening, intralobular reticular opacities, HRCT and AD-CT. U-HRCT images were obtained with a bronchovascular bundle thickening, bronchiectasis, and 1.5-s gantry rotation, 160 mm field of view, 120 kV ,andthree honeycombing), normal anatomical structures (vessels and types of acquisition modes (see Online Supplementary bronchi), and general aspects of image quality (subjective visu- Material): normal mode [U-HRCT : the number of channels al noise, streak artifacts, and dark band artifacts). Overall image per detector row, 896 channels and 0.5 mm × 80 rows; matrix, quality was also evaluated for each image. 512; PF, 0.81; volumetric CT dose index (CTDI ), 23.2 mGy]; Overall image quality, abnormal CT findings, and normal vol super-high-resolution mode (U-HRCT : the number of chan- anatomical structures were subjectively graded using a 3-point SHR nels per detector row, 1792 channels and 0.25 mm × 160 rows; scale: ‘score 1’ indicated poor image quality (i.e., it was pos- matrix, 1024; PF, 0.81; CTDI , 23.3 mGy); volume mode (U- sible to detect structures but difficult to clearly evaluate their vol HRCT : the number of channels per detector row and margin or internal characteristics); ‘score 2’ indicated fair im- SHR-VOL matrix as in U-HRCT ;CTDI , 19.2 mGy). AD-CT images age quality (i.e., the margin or internal characteristics can be SHR vol were obtained with the same parameters as U-HRCT (but with detected and evaluated as well as in the reference images); CTDI , 23.9 mGy). Image data of the whole lungs were ac- ‘score 3’ indicated excellent image quality (i.e., it was easy vol quired on AD-CT and U-HRCT devices with three acquisition to detect findings and to evaluate their margin or internal char- modes, respectively. On U-HRCT images as reference, three acteristics without any indistinct findings). Subjective visual cross-sectional levels with the most conspicuous CT findings noise and artifacts were also graded on a 3-point scale: ‘score were selected from each cadaveric human lung by three chest 1’ indicated strong presence; ‘score 2’ indicated moderate radiologists (A.H., M.Y., and O.H., with 8, 17, and 25 years of presence (i.e., similar to those in the reference images); ‘score experience, respectively) 1 month before starting with the pres- 3’ indicated slight presence or almost absence. On each refer- ent evaluation. Two technologists (A.U. and S.T.) who were not ence image (U-HRCT image), every visual evaluation item involved in image evaluations recorded information on the an- to be scored was indicated using colored markers by two tech- atomical structures (vessels and bronchi), abnormal CT find- nologists (A.U. and S.T.) so that the assessors could identify ings, and artifacts in each cadaveric human lung. We obtained a the position of each visual evaluation item to be evaluated on total of 99 images (33 AD-CT images, 33 U-HRCT images, the other CT images (Fig. 1). SHR and 33 U-HRCT images) for evaluation and 33 SHR-VOL U-HRCT images for reference standard. Both U-HRCT Objective image interpretation N SHR images and U-HRCT images had the same three SHR-VOL cross-sectional levels as U-HRCT images because all image Quantitative image noise measurements were calculated by data were acquired on the same U-HRCT device. AD-CT im- measuring the standard deviation (SD) values in circular re- ages had almost the same three cross-sectional levels as gions of interest (ROI) drawn on the workstation viewer. U-HRCT images. For comparison with AD-CT images with Quantitative image noise measurements were obtained from 0.5 mm thickness, all 132 axial thin-section CT images of air adjacent to the lungs . ROIs (diameter, 20 mm; area 0.5 mm thickness were reconstructed using a lung kernel (FC81) and adaptive iterative dose reduction in three dimen- sions (ADIR 3D). All CT series were anonymized and trans- ferred to a distant workstation viewer by two technologists (A.U. and S.T.) who were not involved in image evaluation. Subjective image interpretation Three independent chest radiologists (A.H., M.Y., and O.H. with 8, 17, and 25 years of experience, respectively) read all 132 images and evaluated them on a 8.3-megapixel, 32-inch color LCD (4K resolution) monitor without prior knowledge of Fig. 1 Evaluation items on reference images. On each reference image histopathological diagnoses or image acquisition parameters. (U-HRCT ), each visual evaluation item to be scored is indicated using colored marks. This U-HRCT image of diffuse alveolar damage shows For each cadaveric human lung, AD-CT, U-HRCT ,and SHR six evaluation items: 1, bronchi; 2, vessels; 3, ground-glass opacity; 4-6, U-HRCT images were evaluated simultaneously in a SHR-VOL interlobular septal thickening. Streak (arrow) and dark band artifacts blinded manner using U-HRCT images as reference. Images N (arrowhead) can also be seen. U-HRCT : ultra-high-resolution CT with normal mode were displayed with a window level of -600 Hounsfield units Eur Radiol (2018) 28:5060–5068 5063 314 mm ) were placed in three homogeneous parts of each tests. Similarly, data from the objective analysis were image and placed in exactly the same location on each select- also analyzed using the Friedman test followed by ed image. Average SDs from these three ROIs were computed post-hoc tests. A p value < 0.05 was considered and compared statistically. significant. Statistical analysis Results All statistical analyses were performed using commer- cially available software: MedCalc version 17.6-64-bit Slit phantom evaluation statistical software (Frank Schoonjans, Mariakerke, Belgium). Median values of the subjective scores of On AD-CT images, only the 0.35-mm slit could be seen clear- the three independent radiologists and the statistical sig- ly. On U-HRCT , all the 0.35-mm, 0.30-mm, and 0.15-mm SHR nificance of any differences among them from the AD- slits could be seen. Therefore, the spatial resolution of AD-CT CT, U-HRCT ,and U-HRCT images were was at least 0.35 mm and that of U-HRCT was at least SHR SHR-VOL SHR assessed using the Friedman test followed by post-hoc 0.15 mm (Fig. 2). Fig. 2 Slit phantom images. Entire picture and layout of the phantom and 20-mm field of view (e). The 0.35-mm slit can be seen in the AD-CT slit phantoms of stainless steel are shown (a). There are four installation image (b) but not clearly with the 0.30-mm slit (c). The 0.15-mm slit sites of slit phantoms. Four stainless steel slit phantoms can be inserted cannot be seen in the AD-CT image (d). However, the 0.15-mm slit can into one installation site at a time (i.e., maximum 16 slit phantoms). In the be seen in the U-HRCT image (e). AD-CT: area detector CT. SHR present study, 0.35-, 0.30-, and 0.15-mm slits were used. AD-CT image U-HRCT : ultra-high-resolution CT with super-high-resolution mode SHR with a 20-mm field of view (b, c,and d)and U-HRCT image with a SHR 5064 Eur Radiol (2018) 28:5060–5068 Subjective evaluation: abnormal CT findings The image quality scores for abnormal CT findings on AD- CT, U-HRCT , and U-HRCT are summarized in SHR SHR-VOL Table 1. Both U-HRCT and U-HRCT abnormal SHR SHR-VOL CT finding scores were significantly higher than those of AD-CT (p < 0.0001) (Fig. 3) except for intralobular reticular opacities. For intralobular reticular opacities, AD-CT abnor- mal CT finding scores were significantly higher than those of U-HRCT and U-HRCT (p < 0.0014) (Fig. 4). SHR SHR-VOL Subjective evaluation: normal anatomical structures and general aspects of image quality The image quality scores for normal anatomical structures and general aspects of image quality on AD-CT, U-HRCT ,and SHR U-HRCT are summarized in Table 2.Both SHR-VOL U-HRCT and U-HRCT normal anatomical struc- SHR SHR-VOL ture scores for bronchi and vessels were significantly higher than those of AD-CT (p < 0.0001) (Fig. 4). Regarding the general aspects of image quality, both U-HCT and SHR U-HRCT indicated significantly decreased streak and SHR-VOL dark band artifacts with respect to AD-CT (p < 0.0001) (Figs. 3 and 4). The subjective visual noise of U-HRCT was SHR-VOL the lowest of the three groups (p < 0.00001). U-HRCT and SHR U-HRCT overall image quality scores were signifi- SHR-VOL cantly higher than those of AD-CT and almost equal to those of U-HRCT (p <0.0001). Quantitative image noise measurements Quantitative noise values (mean ± SD) on CT images were as follows: AD-CT (16.03 ± 5.17), U-HRCT (15.14 ± 5.49), U-HRCT (30.41 ± 4.65), and U-HRCT (26.84 ± SHR SHR-VOL 5.12). Quantitative noise values were significantly higher in the following order: U-HRCT ,U-HRCT ,AD-CT, SHR SHR-VOL and U-HRCT (p <0.0001). Discussion This study showed that ultra-high-resolution CT (U-HRCT SHR and U-HRCT ) significantly improved the visualization SHR-VOL of normal and abnormal CT findings compared with AD-CT, except for intralobular reticular opacities and reduced streak anddarkbandartifacts.U-HRCT and U-HRCT pro- SHR SHR-VOL vided significantly higher overall image quality than AD-CT. In particular, U-HRCT was less noisy than U-HRCT . SHR-VOL SHR The present study is the first to evaluate the image quality of U- HRCT images compared with AD-CT. The use of U-HRCT might enhance image quality by improving spatial resolution, resulting in provision of more detailed information of lung anatomy and pathology. However, our results might be Table 1 Subjective evaluation: abnormal CT findings Abnormal CT Findings Faint nodules Solid nodules GGO Consolidation Emphysema ISP IRO BBT Bronchiectasis Honeycombing N =11 N =26 N =25 N =11 N =12 N =15 N =12 N =12 N =9 N =11 Acquisition mode AD-CT 1.72 ± 0.46*¶ 1.92 ± 0.27*¶ 1.92 ± 0.27*¶ 1.90 ± 0.30*¶ 2.00 ± 0.00*¶ 2.00 ± 0.00*¶ 2.08 ± 0.28*¶ 2.00 ± 0.00*¶ 2.00 ± 0.00*¶ 2.00 ± 0.00*¶ U-HRCT 3.00 ± 0.00* 3.00 ± 0.00* 2.92 ± 0.27* 3.00 ± 0.00* 3.00 ± 0.00* 3.00 ± 0.00* 1.58 ± 0.66* 3.00 ± 0.00* 3.00 ± 0.00* 3.00 ± 0.00* SHR U-HRCT 3.00 ± 0.00¶ 3.00 ± 0.00¶ 2.92 ± 0.27¶ 3.00 ± 0.00¶ 3.00 ± 0.00¶ 3.00 ± 0.00¶ 1.66 ± 0.65¶ 3.00 ± 0.00¶ 3.00 ± 0.00¶ 3.00 ± 0.00¶ SHR-VOL Data are presented as mean ± SD. Data of the subjective image analysis were statistically analyzed using the Friedman test followed by post-hoc tests. GGO = ground-glass opacity, ISP = interlobular septal thickening, IRO = intralobular reticular opacities, BBT = bronchovascular bundle thickening. AD-CT: area detector CT U-HRCT : ultra-high-resolution CT with super-high-resolution mode. SHR U-HRCT : ultra-high-resolution CT with volume mode. SHR-VOL *There was a significant difference between AD-CT and U-HRCT (p < 0.05). SHR ¶There was a significant difference between AD-CT and U-HRCT (p <0.05) SHR-VOL Eur Radiol (2018) 28:5060–5068 5065 Fig. 3 CT images of a cadaveric lung with diffuse panbronchiolitis. h). There are almost no dark band artifacts (f and i). Both U-HRCT (d) SHR Whole and zoomed CT images of AD-CT (a, b,and c), U-HRCT and U-HRCT (g) produced significantly better overall image SHR SHR-VOL (d, e,and f), and U-HRCT (g, h,and i). Tiny (2-mm-diameter) quality than AD-CT (a). AD-CT: area detector CT. U-HRCT :ultra- SHR-VOL SHR nodules show ill-defined margins and unclear internal structure (b). Dark high-resolution CT with super-high-resolution mode U-HRCT : SHR-VOL band artifacts (arrowheads) can be seen (c). Tiny nodules show well- ultra-high-resolution CT with volume mode defined margins and clear internal structure (air bronchiologram) (e and Fig. 4 CT images of a cadaveric lung with pulmonary hemorrhage. AD- such as bronchi and vessels (arrow) can be seen more clearly on CT (area detector CT) image of a cadaveric lung with pulmonary U-HRCT (c) and U-HRCT (d)thanonAD-CT (b). SHR SHR-VOL hemorrhage (a). Zoomed CT images corresponding to a dashed rectangle U-HRCT (c) and U-HRCT (d) seem to also show normal SHR SHR-VOL in (a) are shown (b, c, and d). Intralobular reticular opacities can be anatomical structures such as small bronchi and vessels as low attenuation detected more easily on AD-CT (b)thanonU-HRCT (c)and areas. While dark band artifacts (arrowheads) can be seen in (b), there are SHR U-HRCT (d). On the other hand, normal anatomical structures almost no dark band artifacts in (c and d). SHR-VOL 5066 Eur Radiol (2018) 28:5060–5068 Table 2 Subjective evaluation: normal anatomical structures and general aspects of image quality Normal anatomical structures General aspects of image quality Overall image quality Bronchi Vessels Subjective visual noise Streak artifacts Dark band artifacts N =33 N =29 N =29 N =33 N =33 N =33 Acquisition mode AD-CT 1.93 ± 0.25*¶ 1.86 ± 0.35*¶ 2.00 ± 0.00¶ 2.30 ± 0.68*¶ 2.00 ± 0.00*¶ 2.00 ± 0.00*¶ U-HRCT 3.00 ± 0.00* 3.00 ± 0.00* 2.00 ± 0.35# 2.93 ± 0.34* 3.00 ± 0.00* 3.00 ± 0.00* SHR U-HRCT 3.00 ± 0.00¶ 3.00 ± 0.00¶ 2.39 ± 0.49¶# 3.00 ± 0.00¶ 3.00 ± 0.00¶ 3.00 ± 0.00¶ SHR-VOL Data are presented as mean ± SD. Data of the subjective image analysis were statistically analyzed using the Friedman test followed by post-hoc tests. AD-CT: area detector CT. U-HRCT : ultra-high-resolution CT with super-high-resolution mode. SHR U-HRCT : ultra-high-resolution CT with volume mode. SHR-VOL *There was a significant difference between AD-CT and U-HRCT (p < 0.0001) SHR ¶There was a significant difference between AD-CT and U-HRCT (p < 0.0001) SHR-VOL #There was a significant difference between U-HRCT and U-HRCT (p < 0.0001) SHR SHR-VOL speculative from a clinical point of view, as all data in the size. Further studies of U-HRCT are needed to examine the present study originated from an ex-vivo phantom study effect of the matrix size on image quality. employing explanted lungs imaged free in air. Further analyses In our evaluation of abnormal CT findings in cadaveric are needed to validate our results by using larger cohorts includ- human lungs, both U-HRCT and U-HRCT signif- SHR SHR-VOL ing various diseases in a clinical practice. icantly improved the abnormal CT findings (ground-glass The recent development of U-HRCT technology (i.e., 0.25 opacity, consolidation, emphysema, faint and solid nodules, × 0.25 mm detector element size, a detector with 1792 chan- interlobular septal thickening, bronchiectasis, honeycombing) nels in 160 rows, and 0.4 × 0.5 mm minimum focus size of the compared with AD-CT. A possible reason why these abnor- X-ray tube) improve spatial resolution in both in-plane and mal findings were more conspicuous in U-HRCT could be the body-axis directions. In the present evaluation of a slit phan- increased spatial resolution of the U-HRCT device. Regarding tom using a reconstruction FOVof 20 mm, the spatial resolu- intralobular reticular opacities, however, visual score values tion of U-HRCT (at least 0.15 mm) was at least two times for AD-CT were significantly higher than for U-HRCT SHR higher than that of AD-CT (at least 0.35 mm). In the evalua- and U-HRCT . In general, intralobular reticular opaci- SHR-VOL tion of cadaveric human lungs using a reconstruction FOV of ties refer to the appearance on HRCT of scattered or diffuse 160 mm, the pixel sizes in 512 × 512 (AD-CT) and 1024 × ground-glass attenuation with superimposed interlobular sep- 1024 matrices (U-HRCT and U-HRCT )were tal thickening and intralobular lines [12–14]. This CT finding SHR SHR-VOL 0.313 mm and 0.156 mm, respectively. In this study, ADCT is due to interstitial pulmonary abnormalities and/or alveolar could resolve up to 0.35 mm in spatial resolution (0.313-mm abnormalities [12, 15–17]. We speculate some of the reasons pixel size < 0.35-mm maximum spatial resolution of AD-CT). could be the superior spatial resolution of U-HRCT enables On the other hand, U-HRCT could resolve up to 0.156 mm in visualizing the fine structures around the interlobular septum, spatial resolution (0.156-mm pixel size > 0.15-mm maximum resulting in the relative blurring of linear shadows in a case of spatial resolution of U-HRCT). Therefore, it is important to images including two kinds of abnormal CT findings such as understand the limit of spatial resolution in each CT device intralobular reticular opacities. In other words, fine shadows and confirm the optimal conditions to secure the spatial reso- are recognized as ground-glass attenuation because of a partial lution in a clinical setting. volume effect on AD-CT due to an inferior spatial resolution. Although spatial resolution depends on the matrix size of the As a result, interlobular septal thickening might be relatively reconstructed images, it can never be higher than the maximum conspicuous compared with the surrounding ground-glass at- spatial resolution of the CT device itself. U-HRCT makes it tenuation. Even in those cases that showed the same CT find- possible to reconstruct images with matrixes larger than 512 × ings as on AD-CT, the higher spatial resolution of U-HRCT 512, which is common nowadays . By using a larger matrix might provide further information on the origin of these CT and reducing the pixel size, U-HRCT can provide higher spatial findings. In the future, it will be necessary to correlate the CT resolution for the same FOV size. However, it is important to findings in various diseases with pathological specimens. select a matrix size that is suitable for the intrinsic resolution of In our evaluation of normal anatomical structures and arti- the device, determined by the focus size and the detector element facts in cadaveric human lungs, both U-HRCT and SHR Eur Radiol (2018) 28:5060–5068 5067 U-HRCT significantly improved the visualization of how these factors influence the image quality of U-HRCT. SHR-VOL normal anatomical structures such as bronchi and vessels and Third, image quality was evaluated using CT devices reduced streak and dark band artifacts compared with AD-CT. manufactured by a single company. Presently, however, In the case of abnormal CT findings, the higher spatial resolu- there are no CT devices that offer a similar performance as tion produced by the U-HRCT device can help better visualize the one used in this study. In the future, we expect that U- thin linear opacities such as bronchi and vessels. U-HRCT HRCT devices developed by other companies will be clin- might be able to reduce streak artifacts because the 1792 chan- ically available. Fourth, regarding the slice position of CT nels of the detector element affect the resolution of the X-Y images, we made the utmost effort to get the same cross plane enabling more precise sampling . Moreover, the fre- section images from the different CT devices. However, quency range of the reconstruction algorithm is doubled in U- we could not get exactly the same images. Fifth, the maxi- HRCT as the number of channels is twice that of AD-CT. mum matrix size in this study was 1024, and further analy- Therefore, it is not necessary to forcibly emphasize the high ses are needed to assess the effect of various matrix sizes on spatial frequency region as is done in AD-CT, resulting in al- the image quality of U-HRCT. Sixth, there was a potential most no dark band artifacts caused by an undershoot. source of error concerning the radiation exposure. Radiation Regarding image noise, quantitative noise values were sig- exposure of the volume mode in terms of CTDI was about vol nificantly higher in the following order: U-HRCT , 15% less than the other acquisition modes. Although the SHR U-HRCT , and AD-CT. The subjective visual noise of maximum effort was made so that the CTDI values would SHR-VOL vol U-HRCT , however, was the lowest. This might be due be almost equivalent among the acquisition settings, the SHR-VOL to the visual effects associated with the overall higher image tube current could only be adjusted by 10-mA increments quality of U-HRCT .In this study,AIDR3D, ahybrid on each CT device (i.e., AD-CT, U-HRCT). Ideally, image SHR-VOL iterative reconstruction method, was used for reducing image noise should be evaluated with exactly the same radiation noise. Previous papers have shown that model-based iterative exposure. Moreover, the radiation dose with CTDI higher vol reconstruction imaging could provide higher image quality than 20 mGy used in cadaveric human lungs without chest with lower noise and artifacts [19, 20]. In the future, the com- walls might be high and could not be applied with a similar bined use of model-based iterative reconstruction and U- size-specific dose estimation to humans. In a practical set- HRCT could possibly result in even higher image quality. ting, evaluations of lung image quality on U-HRCT is need- As a whole, U-HRCT and U-HRCT produced a ed under the appropriate radiation dose. Seventh, although SHR SHR-VOL higher overall image quality than AD-CT, almost equal to we evaluated CT findings in pathologically diagnosed ca- U-HRCT , by precisely delineating fine and/or thin structures daveric human lungs, no detailed correlations of CT find- and reducing artifacts. U-HRCT is less noisy than ings with pathological specimens were assessed. SHR-VOL U-HRCT . Regarding the improvements of spatial resolu- Pathological correlation would be needed to investigate SHR tion, Fischbach et al.  demonstrated that thin section im- small-size CT findings, including intralobular reticular ages enhanced resolution, decreased volume averaging from opacities, in the future. Finally, regarding the 11 cadaveric slice to slice, and resulted in an improvement of small nodule human lungs, they were specimens that had been stored in detection, confidence levels, and interobserver agreement. our institution for a long time. Although the 11 lungs were Coche et al.  demonstrated that enhanced multislice spiral inflated and fixed by the Heitzman method and were diag- CT with thin collimation could be used to analyze the nosed by a pathologist, the details (i.e., type of imaging subsegmental pulmonary arteries precisely and might identify technique, acquisition, and reading procedure) were un- even more distal pulmonary arteries. Yoshioka et al. dem- known. We could not consider the influence of the specimen onstrated that U-HRCT with 0.25-mm slices significantly im- fixation method and preservation condition on the patholog- proved the visualization of the artery of Adamkiewicz com- ical diagnosis. At least, however, the conditions of cadaver- pared with 0.5-mm slices. Therefore, in the future, the im- ic human lungs at the time of imaging were the same be- provement of spatial resolution on U-HRCT might have the cause their image data were acquired on both AD-CTand U- possibility to lead to diagnostic imaging advances in the lungs. HRCT devices around almost the same time. Simultaneously, it might be also important to investigate how In conclusion, U-HRCT (U-HRCT and U-HRCT ) SHR SHR-VOL detailed CT findings will affect the patient's outcome. can provide higher image quality than AD-CT by improving the There are several limitations to the study. First, this study spatial resolution and reducing artifacts. U-HRCT is also SHR-VOL included only a small number of cases with a few limited CT more advantageous concerning noise than U-HRCT .U- SHR findings. Second, evaluations of the influence of absorption HRCT may provide more detailed information for lung anatomy and scattering in the human thorax on image quality were and pathology by clearly delineating CT findings (e.g., vessels, lacking. No influence of motion artifacts on image quality bronchi, ground-glass opacity, consolidation, emphysema, faint was evaluated because of the use of cadaveric human lungs. and solid nodules, interlobular septal thickening, bronchiectasis, Therefore, in the future, it will be necessary to investigate and honeycombing). 5068 Eur Radiol (2018) 28:5060–5068 Acknowledgements We thank Mr. Yuya Ito of Canon Medical Systems 5. Yanagawa M, Tomiyama N, Honda O et al (2010) Multidetector CT Corp., Otawara, Japan for his help with scanning the CT phantoms. of the lung: image quality with garnet-based detectors. Radiology 255:944–954 6. Hara T, Urikura A, Ichikawa K et al (2016) Temporal resolution Funding Noriyuki Tomiyama received a research grant from Canon measurement of 128-slice dual source and 320-row area detector Medical Systems Corp. computed tomography scanners in helical acquisition mode using The other authors state that they have not received any funding for this the impulse method. Phys Med 32:625–630 work. 7. Imai Y, Nukui M, Ishihara Y et al (2009) Development and perfor- mance evaluation of an experimental fine pitch detector multislice Compliance with ethical standards CT scanner. Med Phys 36:1120–1127 8. Kakinuma R, Moriyama N, Muramatsu Y et al (2015) Ultra-high- Guarantor The scientific guarantor of this publication is Noriyuki resolution computed tomography of the lung: image quality of a Tomiyama. prototype scanner. PLoS One 10:e0137165 9. Markarian B, Dailey ET (1993) In: Groskin SA (ed) Preparation of Conflict of interest Noriyuki Tomiyama received a research grant from inflated lung specimens. In Heitzman’s The lung: radiologic- Canon Medical Systems Corp., formerly Toshiba Medical Systems. This pathologic correlations, 3rd ed. Mosby, St. Louis, pp 4–12 work was technically supported by Canon Medical Systems Corp. 10. Boehm T, Willmann JK, Hilfiker PR et al (2003) Thin-section CTof Masahiro Yanagawa, Akinori Hata, Osamu Honda, Noriko Kikuchi, the lung: dose electrocardiographic triggering influence diagnosis? and Tomo Miyata have no conflict of interest related to this study. Radiology 229:483–491 11. Goldman LW (2007) Principles of CT and CT technology. J Nucl Statistics and biometry No complex statistical methods were necessary Med Technol 35:115–128 for this paper. 12. Johkoh T, Itoh H, Müller NL et al (1999) Crazy-paving appearance at thin-section CT: spectrum of disease and pathologic findings. Radiology 211:155–160 Informed consent Written informed consent was waived by the Institutional Review Board. 13. Rossi SE, Erasmus JJ, Volpacchio M et al (2003) BCrazy-paving^ pattern at thin-section CT of the lungs: radiologic-pathologic over- view. Radiographics 23:1509–1519 Ethical approval Institutional Review Board approval was obtained. 14. Frazier AA, Franks TJ, Cooke EO et al (2008) From the archives of the AFIP: pulmonary alveolar proteinosis. Radiographics 28:883–899 Study subjects or cohorts overlap Some study subjects or cohorts were 15. Munk PL, Müller NL, Miller RR, Ostrow DN (1988) Pulmonary previously reported in Radiology 2010; 255:944-954 because cadaveric lymphangitic carcinomatosis: CT and pathologic findings. lungs stored in our institution were used in this study. Radiology 166:705–709 16. Naidich DP, Zerhouni EA, Hutchins GM, Genieser NB, McCauley DI, Methodology Siegelman SS (1985) Computed tomography of the pulmonary paren- � prospective chyma. Part 1: Distal air-space disease. J Thorac Imaging 1:39–53 � experimental 17. Leung AN, Miller RR, Müller NL (1993) Parenchymal � performed at one institution opacification in chronic infiltrative lung diseases: CT-pathologic correlation. Radiology 188:209–214 Open Access This article is distributed under the terms of the Creative 18. Richard WW, Muller NL, Naidich DP (2015) High-resolution CT Commons Attribution 4.0 International License (http:// of the lung, fifth edition. Chapter 1. Technical aspects of high- creativecommons.org/licenses/by/4.0/), which permits unrestricted use, resolution CT. Wolters Kuwer. p2-46, distribution, and reproduction in any medium, provided you give 19. Yamada Y, Jinzaki M, Tanami Y et al (2012) Model-based iterative appropriate credit to the original author(s) and the source, provide a link reconstruction technique for ultralow-dose computed tomography to the Creative Commons license, and indicate if changes were made. of the lung: a pilot study. Invest Radiol 47:482–489 20. Yanagawa M, Gyobu T, Leung AN et al (2014) Ultra-low-dose CT of the lung: effect of iterative reconstruction techniques on image quality. Acad Radiol 21:695–703 References 21. Fischbach F, Knollmann F, Griesshaber V, Freund T, Akkol E, Felix R (2003) Detection of pulmonary nodules by multislice computed 1. Hounsfield GN (1973) Computerized transverse axial scanning (to- tomography: improved detection rate with reduced slice thickness. mography). 1. Description of system. Br J Radiol 46:1016–1022 Eur Radiol 13:2378–2383 2. Hu H (1999) Multi-slice helical CT: scan and reconstruction. Med 22. Coche E, Pawlak S, Dechambre S, Maldague B (2003) Peripheral Phys 26:5–18 pulmonary arteries:identification at multi-slice spiral CT with 3D 3. Hurlock GS, Higashino H, Mochizuki T (2009) History of cardiac reconstruction. Eur Radiol 13:815–822 computed tomography: single to 320-detector row multislice com- 23. Yoshioka K, Tanaka R, Takagi H et al (2017) Ultra-high-resolution puted tomography. Int J Cardiovascular Imaging 25:31–42 CT angiography of the artery of Adamkiewicz: a feasibility study. 4. Choi SI, George RT, Schuleri KH et al (2009) Recent developments Neuroradiology. https://doi.org/10.1007/s00234-017-1927-7 in wide-detector cardiac computed tomography. Int J Cardiovascular Imaging 25:23–29
European Radiology – Springer Journals
Published: May 29, 2018
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
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
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.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera