RADIATION DOSE AND IMAGE QUALITY IN PEDIATRIC HEAD CT

RADIATION DOSE AND IMAGE QUALITY IN PEDIATRIC HEAD CT Abstract Our goal was to define a pediatric head CT protocol able to provide images of diagnostic quality, using the least amount of radiation, in children <10 years of age, while using a filtered back projection reconstruction algorithm. Image quality of 119 pediatric head CTs was assessed using a 5-point scoring system. Exams with scores ≥2.5 were considered of sufficient diagnostic quality. The contrast-to-noise ratio (CNR) was also measured. For children <1 year and 1–9 years, all studies performed with CTDIvol ≥ 20.1 mGy (range: 9–46 mGy) and CTDIvol ≥ 27.5 mGy (range: 15–60 mGy) yielded images of diagnostic quality. All diagnostic studies had a minimum CNR of 1.4. These CTDIvol values represent a good balance between image quality and radiation burden. This information can be helpful in designing pediatric head CT protocols with further dose-reduction, namely, iterative reconstruction algorithms and automated exposure control. INTRODUCTION Pediatric CT volume has rapidly increased in the USA due to the introduction of helical CT that allows for fast image acquisition and significantly decreases the need for sedation(1–6). Due to the smaller size of pediatric patients, the effective dose to a child from a given CT study is usually higher than what is received by an adult(4, 7–9). The greater life-time risks in children will likely result in significantly higher lifetime cancer mortality than in adults(7, 10). A retrospective study of a cohort of pediatric patients exposed to CT scans has shown that cumulative doses of 60 mGy to the brain may triple the risk of brain tumor(11). It has been estimated that pediatric CT accounts for ~6–11% of all CT examinations, and a significant percentage of the pediatric scans are conducted in the head and neck region(4, 10, 12–15). A systematic analysis of the relationship between radiation doses and image quality metrics would be beneficial to develop imaging protocols that use the minimum amount of radiation to achieve images of diagnostic quality. The American College of Radiology (ACR) states that, for CT accreditation, the maximum acceptable CTDIvol for head CT in children <1 year of age is 40 mGy, with a diagnostic reference level of 35 mGy(16). In older children, the ACR recommends the application of mAs reduction factors to the standard adult head CT technique(17). We retrospectively evaluated the relationship between the CTDIvol, a measure of the amount of radiation used to perform an examination, subjective image quality, and an objective image quality metric, specifically the contrast-to-noise ratio (CNR). Our purpose was to determine which CT protocols provide the best balance between radiation dose and diagnostic image quality for pediatric head CT in children <10 years of age. MATERIALS AND METHODS Head CT examinations We conducted a departmental quality improvement project to standardize pediatric neuroradiology CT protocol at our institution. The study was approved by the Medical University of South Carolina Institutional Review Board, exempting the study from requiring individual patient’s consent (exemption number 40 363). As part of this project, we retrospectively reviewed all pediatric head CT studies (age range: 0–9 years) performed at our medical center over the course of 6 months. Exclusion criteria were the presence of any devices, e.g. ventriculoperitoneal shunts, EEG leads, external ventricular drains, cochlear implants, inclusion in the field of view of hands holding the patient’s head, and motion artifacts. We included 119 pediatric head CTs obtained over a 6-month period (<1 year, N = 33; 1–9 years, N = 86). Scans were performed on one of the following multidetector row CT scanners: Siemens Sensation 16-slice CT (Siemens Healthcare), Siemens Sensation 64-slice CT (Siemens Healthcare), or Siemens Somatom Definition Flash 128-slice CT (Siemens Healthcare). All examinations were helical acquisitions obtained without the use of automatic tube current modulation. Acquisition parameters, including the kV, mAs, CT dose index (CTDIvol) and DLP were obtained from dose report sheets imported with each study into PACS. The CTDIvol data pertain to the 16-cm acrylic dosimetry phantom. Head CT scans were performed using an image acquisition field of view ranging from 16 to 22 cm. All images were reconstructed using standard filtered back projection (FBP), with a head kernel (H40s for Siemens) and viewed using 3–5 mm thick contiguous axial slices. Demographic data (age and gender) were obtained for each patient. Where patients had more than one head CT, these were treated as separate scans. Data analysis Images were analyzed in a subjective (image quality) and a quantitative (CNR) manner. Subjective evaluation of image quality Four board-certified radiologists, three attending physicians holding a Certificate of Added Qualification in neuroradiology and a neuroradiology fellow, assessed image quality. From each exam, two slices, one at the level of basal ganglia and one at the level of the fourth ventricle, were selected for evaluation (Figure 1)(18). We selected the basal ganglia slice where the internal capsule was best visualized, because it allows an assessment of the differentiation between deep gray matter and white matter structures. We also assessed image quality of the posterior fossa because visualization of the posterior fossa structures can be hindered by scatter noise and beam hardening artifacts. We selected the posterior fossa image where the middle cerebellar peduncles were best visualized. Readers were shown the complete image set on a standard reading room workstation in a randomized manner and were blinded to patient demographics, study indications and protocol parameters during the rating process. Readers were allowed to change window and level settings as would be done during routine image interpretation, and were instructed to assign an integer score ranging from 1 to 5 using the rating system depicted in Table 1. Image evaluation was based on the slices at two locations, the basal ganglia and the posterior fossa(18). For the basal ganglia image, readers were asked to evaluate the differentiation of the putamen, caudate nucleus, thalamus and the internal capsule. For the posterior fossa image, the reader was asked to evaluate the differentiation between gray and white matter, as well as visualization of the brain stem. Figure 1. View largeDownload slide Axial CT image of the head showing placement of ROIs in the gray and white matter of selected regions at the level of the basal ganglia (a) and posterior fossa (b). Please note that ROI analysis was only performed in the region of the basal ganglia and not in the posterior fossa. Figure 1. View largeDownload slide Axial CT image of the head showing placement of ROIs in the gray and white matter of selected regions at the level of the basal ganglia (a) and posterior fossa (b). Please note that ROI analysis was only performed in the region of the basal ganglia and not in the posterior fossa. Table 1. Reader scoring scheme, based on the image quality. Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise View Large Table 1. Reader scoring scheme, based on the image quality. Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise View Large If half of the readers considered the images of ‘satisfactory’ image quality (score = 3) and half of the readers ‘barely satisfactory but answering the study clinical question’ (score = 2), the study was deemed to be of diagnostic image quality. Thus, all studies with average scores ≥2.5 were considered to be of diagnostic quality. We believe that this is a reasonable representation of clinically acceptable image quality for pediatric patients. Inter-rater agreement was assessed as detailed below. Scores of the four readers were averaged for each CT image evaluated, and were plotted as a function of the CTDIvol. Quantitative evaluation of image quality Two circular ROIs of the same size (area ~ 0.18–0.20 cm2) were placed in the gray and white matter at the level of the basal ganglia for each exam with careful exclusion of surrounding structures (Figure 1)(19). Mean Hounsfield units (HU) and standard deviation (SD) were recorded. The HU from the right and left ROIs were averaged. CNR was calculated using the following formula: CNR=(GMmean–WMmean)/SQRT(SD2GM+SD2WM) where, GMmean and WMmean represent the mean gray matter density and white matter density (HU) measured within the ROI. All statistical analyses were performed using IBM Statistical Package for the Social Science Statistics software (SPSS software version 22, Armonk, NY; IBM Corp). Subjective image quality inter-rater agreement between the four readers was assessed using the intraclass correlation coefficient (ICC), which was interpreted as follows: 0–0.2, poor agreement; 0.3–0.4, fair agreement; 0.5–0.6, moderate agreement; 0.7–0.8, strong agreement; >0.8, almost perfect agreement. Spearman’s rank correlation coefficients were used to assess correlation between variables. We also compared CTDIvol, DLP, average image quality, and CNR among different scanners (16-, 64- and 128-slice CT scanners), using Independent-samples Kruskal–Wallis and Mann–Whitney tests. Results were considered statistically significant when p < 0.05. Scatterplots of basal ganglia CNR and CTDIvol as a function of average image quality scores for two age groups (<1 year and 1–9 years) as well as for the pooled data (all ages) were generated. RESULTS Subjective image quality inter-rater agreement was strong for the basal ganglia images (ICC = 0.864, 95% confidence interval = 0.819–0.900, p < 0.001) and for the posterior fossa images (ICC = 0.823, 95% confidence interval = 0.764–0.870, p < 0.001). We found strong correlations between basal ganglia and posterior fossa image quality scores (Spearman’s rho = 0.847, p < 0.001). For this reason, all further analyses were based only on the basal ganglia image data. Table 2 lists the distribution of the CTDIvol, DLP and kV across scanners. Table 3 shows the CTDIvol, DLP and kV used for obtaining the images for the two age groups. We found significant correlation between CTDIvol and average image quality ratings (<1-year-old group: rho = 0.642, p < 0.001; 1–9 years: rho = 0.737, p < 0.001). Figure 2 shows plots of the CNR measured in the basal ganglia as a function of the average image quality scores for the two age groups, as well as the pooled data. Data in Figure 2 allowed us to estimate the CNR corresponding to a minimally acceptable image quality (image quality score ≥ 2.5). Based on the best-fit line to the data, the CNR was calculated to be 1.3 and 1.5 for the 0–1 year and 1–9 years old age groups, respectively. A CNR of 1.4 was calculated from the best-fit line for the pooled data. There was no difference in CNR between studies with acceptable diagnostic quality conducted in infants and 1–9 years old children (total N = 119, p = 0.378, the median [range] CNR in infants was 1.8 [0.7–2.9]; the median [range] CNR in 1–9 years old was 2 [1.1–3.4]). Table 2. Image acquisition parameters for the three CT scanners. The number of studies (n) and the median values ± the standard deviation are listed. n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 Table 2. Image acquisition parameters for the three CT scanners. The number of studies (n) and the median values ± the standard deviation are listed. n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 Table 3. Image acquisition parameters used in study exams. 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 Table 3. Image acquisition parameters used in study exams. 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 Figure 2. View largeDownload slide Contrast-to-noise (CNR) as a function of image quality scores for patients aged (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). From the best-fit line for the pooled data, an image quality score of 2.5 (minimum diagnostically acceptable) corresponds to a CNR of 1.4. Figure 2. View largeDownload slide Contrast-to-noise (CNR) as a function of image quality scores for patients aged (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). From the best-fit line for the pooled data, an image quality score of 2.5 (minimum diagnostically acceptable) corresponds to a CNR of 1.4. Figure 3 shows a plot of the CTDIvol versus image quality scores for the two age groups and for the pooled data. When CTDIvol exceeded 20.1 mGy for infants (less than 1 year) and 27.5 mGy for children from 1 to 9 years of age, 100% of CT images were deemed of adequate diagnostic quality. Figure 3. View largeDownload slide CTDIvol (mGy) versus image quality for the ages (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). Figure 3. View largeDownload slide CTDIvol (mGy) versus image quality for the ages (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). Approximately half of the studies (59/119, 49.6%) were performed using a 64-slice CT scanner, approximately one-third on a 128-slice CT scanner (43/119, 36.1%), and the remaining studies on a 16-slice CT scanner (17/119, 14.2%). CTDIvol, DLP and basal ganglia CNR did not differ among 16-, 64- and 128-slice scanners (respectively, p = 0.109, p = 0.776 and p = 0.225). Subjective image quality differed across scanners (p = 0.005). Post-hoc Mann–Whitney tests showed that image quality was greater for studies done on the 64-slice scanner than the 16-slice scanner (p = 0.005) and 128-slice scanner (p = 0.025), and did not differ between the 16- and 128-slice scanners (p = 0.129). These results are represented in Figure 4. Figure 4. View largeDownload slide Box-plot showing the median image quality score (averaged over all four readers), and range (minimum and maximum as indicated by the whiskers) for the 16-slice (Sensation 16), 64-slice (Sensation 64) and 128-slice CT scanner (Definition Flash). Shown also are the lower and upper quartiles. Figure 4. View largeDownload slide Box-plot showing the median image quality score (averaged over all four readers), and range (minimum and maximum as indicated by the whiskers) for the 16-slice (Sensation 16), 64-slice (Sensation 64) and 128-slice CT scanner (Definition Flash). Shown also are the lower and upper quartiles. DISCUSSION Two strategies should be pursued in pediatric CT imaging. The first is to carefully consider CT scan appropriateness in pediatric patients on a case-by-case basis in light of approved guidelines and to eliminate unnecessary studies(4, 10). In our study the most common indication was trauma (60% of cases), with the remaining 40% being intracranial hemorrhage, seizures, altered mental status, abnormal neurological exam, headache and craniosynostosis. The second strategy is to adjust technical parameters on CT scanners to minimize radiation dose while retaining diagnostic image quality(10, 20, 21). In this study, we evaluated the relationship between the CTDIvol, a measure of the amount of radiation used to perform an examination, subjective image quality and objective image quality (CNR) in pediatric head CT examinations. As expected, we found a strong association between image quality ratings and CTDIvol. This information can be used to design pediatric head CT protocols that minimize patient exposure to ionizing radiation without compromising diagnostic image quality. Optimization of CT protocols is achieved by balancing patient safety and the diagnostic utility of a study(10). When CTDIvol is set too low, the diagnostic performance of a study is compromised because of increased image noise, resulting in the potential for greater false-positive and false-negative rates. In addition, a non-diagnostic study may need to be repeated, which leads to an overall increase of patient’s radiation dose. It should be noted that the use of too little radiation is particularly detrimental in pediatric head CT imaging compared to other pediatric CT studies, because subtle differences in brain attenuation can be easily obscured by excessive image noise. As a result, for head CT the mAs is typically decreased only by a factor of up to 2–2.5 from an adult to a newborn, while for CT imaging of the body, a reduction in mAs by a factor of up to 4–5 is applied(22). Our results support a more aggressive dose reduction than recommended by the American College of Radiology (ACR) for head CT in infants. In fact, the ACR CT accreditation standards state that the maximum acceptable CTDIvol for head CT in infants is 40 mGy, with a reference level of 35 mGy(16). A nationwide survey of radiation doses for pediatric head CT examinations in children <12 months of age revealed that average reported CTDIvol was 27.3 mGy for all hospitals (including pediatric facilities) and 22.3 mGy for pediatric specialty hospitals(23). Of note, corresponding image quality data were not available or reported in this study. Our results are in line with the findings of this survey. In fact, we found that when CTDIvol exceeded ~20 mGy for infants, 100% of CT images were of diagnostic quality for routine head CT exams performed with multidetector systems and filtered back projection reconstruction. Thus, we feel that although the ACR recommendations set a maximum CTDIvol of 40 mGy for head CT in infants, the amount of radiation used in this age group can be further decreased, while preserving diagnostic image quality. We have also found that a CTDIvol of at least 28 mGy resulted in CT images of adequate quality in 100% of examinations in the 1–9 years old group. Although the ACR does not provide CT accreditation standards for head CT in older children, mAs reduction factors and protocol recommendations are readily available(17). According to the Image Gently guidelines, if a routine head CT examination in an adult is conducted at 120 kV, 320 mAs and a CTDIvol of 60 mGy, then a head CT in a 2-year-old and in a 6-year-old child could be performed respectively at 120 kV, 275 mAs and 298 mAs (CTDIvol ~ 52 mGy and 56 mGy), for a limited mAs reduction. For a moderate mAs reduction, the recommended values in a 2-year-old and in a 6-year-old are respectively 198 and 253 mAs (CTDIvol ~ 37 and 47 mGy). Our results show that more aggressive dose reduction may also be pursued in children aged 1–9 without compromising the diagnostic image quality. The CNR in our study sample (average CNR = 1.90 ± 0.55) was similar to the CNR of standard head CT examinations in adult patients (average CNR = 1.77 ± 0.52) reported by Mullins et al.(24), and to the CNR for standard (average CNR = 1.47 ± 0.28) and dual-energy head CT (average CNR = 1.97 ± 0.50) in adults reported by Pomerantz et al.(25) In our study, plots of the CNR measured in the basal ganglia as a function of the average image quality scores demonstrated that a CNR of 1.4 corresponds to image quality sufficient for adequate interpretation. CNR of at least 1.4 resulted in images of adequate quality in all infant head CT, and ~98.5% of cases in the 1–9 years group. It is interesting to note that CNR values associated with images of diagnostic quality were similar across the two pediatric age groups evaluated in this study. Head CT exams were performed using 16-, 64- and 128-slice CT scanners. We explored differences in technique, image quality ratings, and CNR among scanners, although we may not have had enough power to fully characterize differences in performance. We did not find significant differences in CTDIvol, DLP and CNR among scanners. However, studies conducted using the 64-slice CT scanner had higher quality ratings than those performed using the 16-slice CT scanner. We attribute this to the fact that the 64-slice scanner was newer, with better reconstruction filters and detector technology as compared to the 16-slice scanner. This retrospective study has several limitations. The study scans were conducted using 16-, 64- and 128-slice multidetector CT scanners. While we believe that this is a reasonable representation of most CT scanners currently in use, these data may not be necessarily representative of scans performed with other CT scanners with dose reduction software and/or detectors. Furthermore, we evaluated objective and subjective image quality of studies performed using a fixed tube current and tube potential; therefore, our results may not be applicable to head CT examinations performed using tube voltage and tube current modulation techniques. In this study CT reconstructions were performed without the use of iterative reconstruction algorithms, which have the potential to allow reduction in radiation dose while preserving image quality compared with conventional filtered back projection(26). Further, it should be noted that our results are preliminary and should be confirmed in future studies with larger sample size. The wide range of dose indices (Table 2) reported for the same CT procedure and age group underscores the need for standardization and can be effectively addressed by the implementation of standardized protocols and continued education of radiology staff(10, 19, 27, 28). CONCLUSION In conclusion, when CTDIvol exceeded ~20 mGy for infants (<1 year) and 28 mGy for children from 1 to 9 years of age, 100% of CT images were deemed of diagnostic quality for routine head CT exams performed with multidetector systems. We believe these CTDIvol values represent a good balance between image quality and radiation burden. At our institution, we have adopted a standardized approach to pediatric head CT imaging based on the results of this study. Additionally, continued efforts are underway to optimize head CT imaging in pediatric patients. Our results pertain to head CT protocols with filtered back projection algorithm, and further studies should evaluate the relationship between CTDIvol and image quality for head CT protocols that employ iterative reconstruction and dose modulation techniques. IRB STATEMENT An appropriate institutional review board approved the study. GRANT SUPPORT None. PRESENTATION AT MEETINGS ASNR Meeting 2016, ‘Optimizing the Balance between Radiation Dose and Image Quality in Pediatric Head CT’. Abstract No: O-74. REFERENCES 1 Broder , J. , Fordham , L. A. and Warshauer , D. M. Increasing utilization of computed tomography in the pediatric emergency department, 2000–2006 . Emerg. Radiol. 14 ( 4 ), 227 – 232 ( 2007 ). Google Scholar CrossRef Search ADS PubMed 2 Markel , T. A. , Kumar , R. , Koontz , N. A. , Scherer , L. R. and Applegate , K. E. The utility of computed tomography as a screening tool for the evaluation of pediatric blunt chest trauma . J. Trauma 67 ( 1 ), 23 – 28 ( 2009 ). Google Scholar CrossRef Search ADS PubMed 3 Blackwell , C. D. , Gorelick , M. , Holmes , J. F. , Bandyopadhyay , S. and Kuppermann , N. Pediatric head trauma: changes in use of computed tomography in emergency departments in the United States over time . Ann. Emerg. Med. 49 ( 3 ), 320 – 324 ( 2007 ). Google Scholar CrossRef Search ADS PubMed 4 Brenner , D. J. and Hall , E. J. Computed tomography—an increasing source of radiation exposure . N. Engl. J. Med. 357 ( 22 ), 2277 – 2284 ( 2007 ). Google Scholar CrossRef Search ADS PubMed 5 White , K. S. Reduced need for sedation in patients undergoing helical CT of the chest and abdomen . Pediatr. Radiol. 25 ( 5 ), 344 – 346 ( 1995 ). Google Scholar CrossRef Search ADS PubMed 6 Shahi , V. , Brinjikji , W. , Cloft , H. J. , Thomas , K. B. and Kallmes , D. F. Trends in CT utilization for pediatric fall patients in US Emergency Departments . Acad. Radiol. 22 ( 7 ), 898 – 903 ( 2015 ). Google Scholar CrossRef Search ADS PubMed 7 Brenner , D. , Elliston , C. , Hall , E. and Berdon , W. Estimated risks of radiation-induced fatal cancer from pediatric CT . AJR Am. J. Roentgenol. 176 ( 2 ), 289 – 296 ( 2001 ). Google Scholar CrossRef Search ADS PubMed 8 United Nations Scientific Committee on the Effects of Atomic Radiation. Sources, Effects and Risks of Ionizing Radiation. I ( 2013 ). http://www.unscear.org/docs/reports/2013/13-85418_Report_2013_Annex_A.pdf 9 Brenner , D. J. Estimating cancer risks from pediatric CT: going from the qualitative to the quantitative . Pediatr. Radiol. 32 ( 4 ), 228 – 231 ( 2002 ) ; discussion 242–4. Google Scholar CrossRef Search ADS PubMed 10 Miglioretti , D. L. et al. . The use of computed tomography in pediatrics and the associated radiation exposure and estimated cancer risk . JAMA Pediatr. 167 ( 8 ), 700 – 707 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 11 Pearce , M. S. et al. . Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study . Lancet 380 ( 9840 ), 499 – 505 ( 2012 ). Google Scholar CrossRef Search ADS PubMed 12 Mettler , F. A. , Jr. , Wiest , P. W. , Locken , J. A. and Kelsey , C. A. CT scanning: patterns of use and dose . J. Radiol. Prot. 20 ( 4 ), 353 – 359 ( 2000 ). Google Scholar CrossRef Search ADS PubMed 13 Yamauchi-Kawaura , C. , Fujii , K. , Aoyama , T. , Koyama , S. and Yamauchi , M. Radiation dose evaluation in head and neck MDCT examinations with a 6-year-old child anthropomorphic phantom . Pediatr. Radiol. 40 ( 7 ), 1206 – 1214 ( 2010 ). Google Scholar CrossRef Search ADS PubMed 14 Zacharias , C. , Alessio , A. M. , Otto , R. K. , Iyer , R. S. , Philips , G. S. , Swanson , J. O. and Thapa , M. M. Pediatric CT: strategies to lower radiation dose . AJR Am. J. Roentgenol. 200 ( 5 ), 950 – 956 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 15 Berrington de Gonzalez , A. , Mahesh , M. , Kim , K. P. , Bhargavan , M. , Lewis , R. , Mettler , F. and Land , C. Projected cancer risks from computed tomographic scans performed in the United States in 2007 . Arch. Intern. Med. 169 ( 22 ), 2071 – 2077 ( 2009 ). Google Scholar CrossRef Search ADS PubMed 16 American College of Radiology. Accreditation Program Requirements ( 2016 ). Available on http://www.acraccreditation.org/~/media/ACRAccreditation/Documents/CT/Requirements.pdf?la=en. 17 The Alliance for Radiation Safety in Pediatric Imaging. Image Gently: Computed Tomography ( 2014 ). Available on http://www.imagegently.org/Procedures/Interventional-Radiology/Protocols. 18 Tipnis , S. , Thampy , R. , Rumboldt , Z. , Spampinato , M. , Matheus , G. and Huda , W. Radiation intensity (CTDIvol) and visibility of anatomical structures in head CT examinations . J. Appl. Clin. Med. Phys. 17 ( 1 ), 293 – 300 ( 2016 ). Google Scholar CrossRef Search ADS PubMed 19 Paolicchi , F. , Faggioni , L. , Bastiani , L. , Molinaro , S. , Puglioli , M. , Caramella , D. and Bartolozzi , C. Optimizing the balance between radiation dose and image quality in pediatric head CT: findings before and after intensive radiologic staff training . AJR Am. J. Roentgenol. 202 ( 6 ), 1309 – 1315 ( 2014 ). Google Scholar CrossRef Search ADS PubMed 20 Linton , O. W. and Mettler , F. A. , Jr. National conference on dose reduction in CT, with an emphasis on pediatric patients . AJR Am. J. Roentgenol. 181 ( 2 ), 321 – 329 ( 2003 ). Google Scholar CrossRef Search ADS PubMed 21 Lee , C. , Pearce , M. S. , Salotti , J. A. , Harbron , R. W. , Little , M. P. , McHugh , K. , Chapple , C. L. and Berrington de Gonzalez , A. Reduction in radiation doses from paediatric CT scans in Great Britain . Br. J. Radiol. 89 ( 1060 ), 20150305 ( 2016 ). Google Scholar CrossRef Search ADS PubMed 22 McCollough , C. H. , Primak , A. N. , Braun , N. , Kofler , J. , Yu , L. and Christner , J. Strategies for reducing radiation dose in CT . Radiol. Clin. North Am. 47 ( 1 ), 27 – 40 ( 2009 ). Google Scholar CrossRef Search ADS PubMed 23 Kanal , K. M. , Graves , J. M. , Vavilala , M. S. , Applegate , K. E. , Jarvik , J. G. and Rivara , F. P. Variation in CT pediatric head examination radiation dose: results from a national survey . AJR Am. J. Roentgenol. 204 ( 3 ), W293 – W301 ( 2015 ). Google Scholar CrossRef Search ADS PubMed 24 Mullins , M. E. , Lev , M. H. , Bove , P. , O’Reilly , C. E. , Saini , S. , Rhea , J. T. , Thrall , J. H. , Hunter , G. J. , Hamberg , L. M. and Gonzalez , R. G. Comparison of image quality between conventional and low-dose nonenhanced head CT . AJNR Am. J. Neuroradiol. 25 ( 4 ), 533 – 538 ( 2004 ). Google Scholar PubMed 25 Pomerantz , S. R. , Kamalian , S. , Zhang , D. , Gupta , R. , Rapalino , O. , Sahani , D. V. and Lev , M. H. Virtual monochromatic reconstruction of dual-energy unenhanced head CT at 65-75 keV maximizes image quality compared with conventional polychromatic CT . Radiology 266 ( 1 ), 318 – 325 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 26 Kilic , K. , Erbas , G. , Guryildirim , M. , Konus , O. L. , Arac , M. , Ilgit , E. and Isik , S. Quantitative and qualitative comparison of standard-dose and low-dose pediatric head computed tomography: a retrospective study assessing the effect of adaptive statistical iterative reconstruction . J. Comput. Assist. Tomogr. 37 ( 3 ), 377 – 381 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 27 Goske , M. J. et al. . The Image Gently campaign: working together to change practice . AJR Am. J. Roentgenol. 190 ( 2 ), 273 – 274 ( 2008 ). Google Scholar CrossRef Search ADS PubMed 28 De Bondt , T. , Mulkens , T. , Zanca , F. , Pyfferoen , L. , Casselman , J. W. and Parizel , P. M. Benchmarking pediatric cranial CT protocols using a dose tracking software system: a multicenter study . Eur. Radiol. 27 ( 2 ), 841 – 850 ( 2017 ). Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Radiation Protection Dosimetry Oxford University Press

Loading next page...
 
/lp/ou_press/radiation-dose-and-image-quality-in-pediatric-head-ct-u0NemswD0W
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com
ISSN
0144-8420
eISSN
1742-3406
D.O.I.
10.1093/rpd/ncy066
Publisher site
See Article on Publisher Site

Abstract

Abstract Our goal was to define a pediatric head CT protocol able to provide images of diagnostic quality, using the least amount of radiation, in children <10 years of age, while using a filtered back projection reconstruction algorithm. Image quality of 119 pediatric head CTs was assessed using a 5-point scoring system. Exams with scores ≥2.5 were considered of sufficient diagnostic quality. The contrast-to-noise ratio (CNR) was also measured. For children <1 year and 1–9 years, all studies performed with CTDIvol ≥ 20.1 mGy (range: 9–46 mGy) and CTDIvol ≥ 27.5 mGy (range: 15–60 mGy) yielded images of diagnostic quality. All diagnostic studies had a minimum CNR of 1.4. These CTDIvol values represent a good balance between image quality and radiation burden. This information can be helpful in designing pediatric head CT protocols with further dose-reduction, namely, iterative reconstruction algorithms and automated exposure control. INTRODUCTION Pediatric CT volume has rapidly increased in the USA due to the introduction of helical CT that allows for fast image acquisition and significantly decreases the need for sedation(1–6). Due to the smaller size of pediatric patients, the effective dose to a child from a given CT study is usually higher than what is received by an adult(4, 7–9). The greater life-time risks in children will likely result in significantly higher lifetime cancer mortality than in adults(7, 10). A retrospective study of a cohort of pediatric patients exposed to CT scans has shown that cumulative doses of 60 mGy to the brain may triple the risk of brain tumor(11). It has been estimated that pediatric CT accounts for ~6–11% of all CT examinations, and a significant percentage of the pediatric scans are conducted in the head and neck region(4, 10, 12–15). A systematic analysis of the relationship between radiation doses and image quality metrics would be beneficial to develop imaging protocols that use the minimum amount of radiation to achieve images of diagnostic quality. The American College of Radiology (ACR) states that, for CT accreditation, the maximum acceptable CTDIvol for head CT in children <1 year of age is 40 mGy, with a diagnostic reference level of 35 mGy(16). In older children, the ACR recommends the application of mAs reduction factors to the standard adult head CT technique(17). We retrospectively evaluated the relationship between the CTDIvol, a measure of the amount of radiation used to perform an examination, subjective image quality, and an objective image quality metric, specifically the contrast-to-noise ratio (CNR). Our purpose was to determine which CT protocols provide the best balance between radiation dose and diagnostic image quality for pediatric head CT in children <10 years of age. MATERIALS AND METHODS Head CT examinations We conducted a departmental quality improvement project to standardize pediatric neuroradiology CT protocol at our institution. The study was approved by the Medical University of South Carolina Institutional Review Board, exempting the study from requiring individual patient’s consent (exemption number 40 363). As part of this project, we retrospectively reviewed all pediatric head CT studies (age range: 0–9 years) performed at our medical center over the course of 6 months. Exclusion criteria were the presence of any devices, e.g. ventriculoperitoneal shunts, EEG leads, external ventricular drains, cochlear implants, inclusion in the field of view of hands holding the patient’s head, and motion artifacts. We included 119 pediatric head CTs obtained over a 6-month period (<1 year, N = 33; 1–9 years, N = 86). Scans were performed on one of the following multidetector row CT scanners: Siemens Sensation 16-slice CT (Siemens Healthcare), Siemens Sensation 64-slice CT (Siemens Healthcare), or Siemens Somatom Definition Flash 128-slice CT (Siemens Healthcare). All examinations were helical acquisitions obtained without the use of automatic tube current modulation. Acquisition parameters, including the kV, mAs, CT dose index (CTDIvol) and DLP were obtained from dose report sheets imported with each study into PACS. The CTDIvol data pertain to the 16-cm acrylic dosimetry phantom. Head CT scans were performed using an image acquisition field of view ranging from 16 to 22 cm. All images were reconstructed using standard filtered back projection (FBP), with a head kernel (H40s for Siemens) and viewed using 3–5 mm thick contiguous axial slices. Demographic data (age and gender) were obtained for each patient. Where patients had more than one head CT, these were treated as separate scans. Data analysis Images were analyzed in a subjective (image quality) and a quantitative (CNR) manner. Subjective evaluation of image quality Four board-certified radiologists, three attending physicians holding a Certificate of Added Qualification in neuroradiology and a neuroradiology fellow, assessed image quality. From each exam, two slices, one at the level of basal ganglia and one at the level of the fourth ventricle, were selected for evaluation (Figure 1)(18). We selected the basal ganglia slice where the internal capsule was best visualized, because it allows an assessment of the differentiation between deep gray matter and white matter structures. We also assessed image quality of the posterior fossa because visualization of the posterior fossa structures can be hindered by scatter noise and beam hardening artifacts. We selected the posterior fossa image where the middle cerebellar peduncles were best visualized. Readers were shown the complete image set on a standard reading room workstation in a randomized manner and were blinded to patient demographics, study indications and protocol parameters during the rating process. Readers were allowed to change window and level settings as would be done during routine image interpretation, and were instructed to assign an integer score ranging from 1 to 5 using the rating system depicted in Table 1. Image evaluation was based on the slices at two locations, the basal ganglia and the posterior fossa(18). For the basal ganglia image, readers were asked to evaluate the differentiation of the putamen, caudate nucleus, thalamus and the internal capsule. For the posterior fossa image, the reader was asked to evaluate the differentiation between gray and white matter, as well as visualization of the brain stem. Figure 1. View largeDownload slide Axial CT image of the head showing placement of ROIs in the gray and white matter of selected regions at the level of the basal ganglia (a) and posterior fossa (b). Please note that ROI analysis was only performed in the region of the basal ganglia and not in the posterior fossa. Figure 1. View largeDownload slide Axial CT image of the head showing placement of ROIs in the gray and white matter of selected regions at the level of the basal ganglia (a) and posterior fossa (b). Please note that ROI analysis was only performed in the region of the basal ganglia and not in the posterior fossa. Table 1. Reader scoring scheme, based on the image quality. Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise View Large Table 1. Reader scoring scheme, based on the image quality. Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise Reader score Interpretation Comment 1 Unacceptable Poor image quality, study is non-diagnostic 2 Barely satisfactory Poor image quality, but answers the major clinical questions (Mass? Hemorrhage? Clear infarction?) 3 Satisfactory Image quality sufficient for adequate interpretation, however with artifacts and/or noise, may obscure very subtle details 4 Good Above average image quality with the noise and artifacts not affecting diagnostic value 5 Excellent Excellent image quality, free of artifacts and with imperceptible noise View Large If half of the readers considered the images of ‘satisfactory’ image quality (score = 3) and half of the readers ‘barely satisfactory but answering the study clinical question’ (score = 2), the study was deemed to be of diagnostic image quality. Thus, all studies with average scores ≥2.5 were considered to be of diagnostic quality. We believe that this is a reasonable representation of clinically acceptable image quality for pediatric patients. Inter-rater agreement was assessed as detailed below. Scores of the four readers were averaged for each CT image evaluated, and were plotted as a function of the CTDIvol. Quantitative evaluation of image quality Two circular ROIs of the same size (area ~ 0.18–0.20 cm2) were placed in the gray and white matter at the level of the basal ganglia for each exam with careful exclusion of surrounding structures (Figure 1)(19). Mean Hounsfield units (HU) and standard deviation (SD) were recorded. The HU from the right and left ROIs were averaged. CNR was calculated using the following formula: CNR=(GMmean–WMmean)/SQRT(SD2GM+SD2WM) where, GMmean and WMmean represent the mean gray matter density and white matter density (HU) measured within the ROI. All statistical analyses were performed using IBM Statistical Package for the Social Science Statistics software (SPSS software version 22, Armonk, NY; IBM Corp). Subjective image quality inter-rater agreement between the four readers was assessed using the intraclass correlation coefficient (ICC), which was interpreted as follows: 0–0.2, poor agreement; 0.3–0.4, fair agreement; 0.5–0.6, moderate agreement; 0.7–0.8, strong agreement; >0.8, almost perfect agreement. Spearman’s rank correlation coefficients were used to assess correlation between variables. We also compared CTDIvol, DLP, average image quality, and CNR among different scanners (16-, 64- and 128-slice CT scanners), using Independent-samples Kruskal–Wallis and Mann–Whitney tests. Results were considered statistically significant when p < 0.05. Scatterplots of basal ganglia CNR and CTDIvol as a function of average image quality scores for two age groups (<1 year and 1–9 years) as well as for the pooled data (all ages) were generated. RESULTS Subjective image quality inter-rater agreement was strong for the basal ganglia images (ICC = 0.864, 95% confidence interval = 0.819–0.900, p < 0.001) and for the posterior fossa images (ICC = 0.823, 95% confidence interval = 0.764–0.870, p < 0.001). We found strong correlations between basal ganglia and posterior fossa image quality scores (Spearman’s rho = 0.847, p < 0.001). For this reason, all further analyses were based only on the basal ganglia image data. Table 2 lists the distribution of the CTDIvol, DLP and kV across scanners. Table 3 shows the CTDIvol, DLP and kV used for obtaining the images for the two age groups. We found significant correlation between CTDIvol and average image quality ratings (<1-year-old group: rho = 0.642, p < 0.001; 1–9 years: rho = 0.737, p < 0.001). Figure 2 shows plots of the CNR measured in the basal ganglia as a function of the average image quality scores for the two age groups, as well as the pooled data. Data in Figure 2 allowed us to estimate the CNR corresponding to a minimally acceptable image quality (image quality score ≥ 2.5). Based on the best-fit line to the data, the CNR was calculated to be 1.3 and 1.5 for the 0–1 year and 1–9 years old age groups, respectively. A CNR of 1.4 was calculated from the best-fit line for the pooled data. There was no difference in CNR between studies with acceptable diagnostic quality conducted in infants and 1–9 years old children (total N = 119, p = 0.378, the median [range] CNR in infants was 1.8 [0.7–2.9]; the median [range] CNR in 1–9 years old was 2 [1.1–3.4]). Table 2. Image acquisition parameters for the three CT scanners. The number of studies (n) and the median values ± the standard deviation are listed. n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 Table 2. Image acquisition parameters for the three CT scanners. The number of studies (n) and the median values ± the standard deviation are listed. n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 n CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Sensation 16 16 35 ± 16 663 ± 325 99 ± 17 Sensation 64 59 41 ± 13 677 ± 244 111 ± 10 Definition Flash 44 39 ± 14 647 ± 270 107 ± 10 Table 3. Image acquisition parameters used in study exams. 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 Table 3. Image acquisition parameters used in study exams. 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 0–1 y (n = 33) CTDIvol (mGy) DLP (mGy cm) Voltage (kV) Min 9 130 80 25%ile 24 349 100 Median 37 574 100 Std. dev. 12 193 8 75%ile 45 664 100 Max 46 788 120 1–9 y (n = 86) CTDIvol (mGy) DLP (mGy cm) kV Min 15 213 80 25%ile 24 450 100 Median 47 778 120 Std. dev. 14 264 12 75%ile 55 965 120 Max 60 1167 120 Figure 2. View largeDownload slide Contrast-to-noise (CNR) as a function of image quality scores for patients aged (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). From the best-fit line for the pooled data, an image quality score of 2.5 (minimum diagnostically acceptable) corresponds to a CNR of 1.4. Figure 2. View largeDownload slide Contrast-to-noise (CNR) as a function of image quality scores for patients aged (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). From the best-fit line for the pooled data, an image quality score of 2.5 (minimum diagnostically acceptable) corresponds to a CNR of 1.4. Figure 3 shows a plot of the CTDIvol versus image quality scores for the two age groups and for the pooled data. When CTDIvol exceeded 20.1 mGy for infants (less than 1 year) and 27.5 mGy for children from 1 to 9 years of age, 100% of CT images were deemed of adequate diagnostic quality. Figure 3. View largeDownload slide CTDIvol (mGy) versus image quality for the ages (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). Figure 3. View largeDownload slide CTDIvol (mGy) versus image quality for the ages (a) 0–1 y, (b) 1–9 y and (c) 0–9 y (pooled). Approximately half of the studies (59/119, 49.6%) were performed using a 64-slice CT scanner, approximately one-third on a 128-slice CT scanner (43/119, 36.1%), and the remaining studies on a 16-slice CT scanner (17/119, 14.2%). CTDIvol, DLP and basal ganglia CNR did not differ among 16-, 64- and 128-slice scanners (respectively, p = 0.109, p = 0.776 and p = 0.225). Subjective image quality differed across scanners (p = 0.005). Post-hoc Mann–Whitney tests showed that image quality was greater for studies done on the 64-slice scanner than the 16-slice scanner (p = 0.005) and 128-slice scanner (p = 0.025), and did not differ between the 16- and 128-slice scanners (p = 0.129). These results are represented in Figure 4. Figure 4. View largeDownload slide Box-plot showing the median image quality score (averaged over all four readers), and range (minimum and maximum as indicated by the whiskers) for the 16-slice (Sensation 16), 64-slice (Sensation 64) and 128-slice CT scanner (Definition Flash). Shown also are the lower and upper quartiles. Figure 4. View largeDownload slide Box-plot showing the median image quality score (averaged over all four readers), and range (minimum and maximum as indicated by the whiskers) for the 16-slice (Sensation 16), 64-slice (Sensation 64) and 128-slice CT scanner (Definition Flash). Shown also are the lower and upper quartiles. DISCUSSION Two strategies should be pursued in pediatric CT imaging. The first is to carefully consider CT scan appropriateness in pediatric patients on a case-by-case basis in light of approved guidelines and to eliminate unnecessary studies(4, 10). In our study the most common indication was trauma (60% of cases), with the remaining 40% being intracranial hemorrhage, seizures, altered mental status, abnormal neurological exam, headache and craniosynostosis. The second strategy is to adjust technical parameters on CT scanners to minimize radiation dose while retaining diagnostic image quality(10, 20, 21). In this study, we evaluated the relationship between the CTDIvol, a measure of the amount of radiation used to perform an examination, subjective image quality and objective image quality (CNR) in pediatric head CT examinations. As expected, we found a strong association between image quality ratings and CTDIvol. This information can be used to design pediatric head CT protocols that minimize patient exposure to ionizing radiation without compromising diagnostic image quality. Optimization of CT protocols is achieved by balancing patient safety and the diagnostic utility of a study(10). When CTDIvol is set too low, the diagnostic performance of a study is compromised because of increased image noise, resulting in the potential for greater false-positive and false-negative rates. In addition, a non-diagnostic study may need to be repeated, which leads to an overall increase of patient’s radiation dose. It should be noted that the use of too little radiation is particularly detrimental in pediatric head CT imaging compared to other pediatric CT studies, because subtle differences in brain attenuation can be easily obscured by excessive image noise. As a result, for head CT the mAs is typically decreased only by a factor of up to 2–2.5 from an adult to a newborn, while for CT imaging of the body, a reduction in mAs by a factor of up to 4–5 is applied(22). Our results support a more aggressive dose reduction than recommended by the American College of Radiology (ACR) for head CT in infants. In fact, the ACR CT accreditation standards state that the maximum acceptable CTDIvol for head CT in infants is 40 mGy, with a reference level of 35 mGy(16). A nationwide survey of radiation doses for pediatric head CT examinations in children <12 months of age revealed that average reported CTDIvol was 27.3 mGy for all hospitals (including pediatric facilities) and 22.3 mGy for pediatric specialty hospitals(23). Of note, corresponding image quality data were not available or reported in this study. Our results are in line with the findings of this survey. In fact, we found that when CTDIvol exceeded ~20 mGy for infants, 100% of CT images were of diagnostic quality for routine head CT exams performed with multidetector systems and filtered back projection reconstruction. Thus, we feel that although the ACR recommendations set a maximum CTDIvol of 40 mGy for head CT in infants, the amount of radiation used in this age group can be further decreased, while preserving diagnostic image quality. We have also found that a CTDIvol of at least 28 mGy resulted in CT images of adequate quality in 100% of examinations in the 1–9 years old group. Although the ACR does not provide CT accreditation standards for head CT in older children, mAs reduction factors and protocol recommendations are readily available(17). According to the Image Gently guidelines, if a routine head CT examination in an adult is conducted at 120 kV, 320 mAs and a CTDIvol of 60 mGy, then a head CT in a 2-year-old and in a 6-year-old child could be performed respectively at 120 kV, 275 mAs and 298 mAs (CTDIvol ~ 52 mGy and 56 mGy), for a limited mAs reduction. For a moderate mAs reduction, the recommended values in a 2-year-old and in a 6-year-old are respectively 198 and 253 mAs (CTDIvol ~ 37 and 47 mGy). Our results show that more aggressive dose reduction may also be pursued in children aged 1–9 without compromising the diagnostic image quality. The CNR in our study sample (average CNR = 1.90 ± 0.55) was similar to the CNR of standard head CT examinations in adult patients (average CNR = 1.77 ± 0.52) reported by Mullins et al.(24), and to the CNR for standard (average CNR = 1.47 ± 0.28) and dual-energy head CT (average CNR = 1.97 ± 0.50) in adults reported by Pomerantz et al.(25) In our study, plots of the CNR measured in the basal ganglia as a function of the average image quality scores demonstrated that a CNR of 1.4 corresponds to image quality sufficient for adequate interpretation. CNR of at least 1.4 resulted in images of adequate quality in all infant head CT, and ~98.5% of cases in the 1–9 years group. It is interesting to note that CNR values associated with images of diagnostic quality were similar across the two pediatric age groups evaluated in this study. Head CT exams were performed using 16-, 64- and 128-slice CT scanners. We explored differences in technique, image quality ratings, and CNR among scanners, although we may not have had enough power to fully characterize differences in performance. We did not find significant differences in CTDIvol, DLP and CNR among scanners. However, studies conducted using the 64-slice CT scanner had higher quality ratings than those performed using the 16-slice CT scanner. We attribute this to the fact that the 64-slice scanner was newer, with better reconstruction filters and detector technology as compared to the 16-slice scanner. This retrospective study has several limitations. The study scans were conducted using 16-, 64- and 128-slice multidetector CT scanners. While we believe that this is a reasonable representation of most CT scanners currently in use, these data may not be necessarily representative of scans performed with other CT scanners with dose reduction software and/or detectors. Furthermore, we evaluated objective and subjective image quality of studies performed using a fixed tube current and tube potential; therefore, our results may not be applicable to head CT examinations performed using tube voltage and tube current modulation techniques. In this study CT reconstructions were performed without the use of iterative reconstruction algorithms, which have the potential to allow reduction in radiation dose while preserving image quality compared with conventional filtered back projection(26). Further, it should be noted that our results are preliminary and should be confirmed in future studies with larger sample size. The wide range of dose indices (Table 2) reported for the same CT procedure and age group underscores the need for standardization and can be effectively addressed by the implementation of standardized protocols and continued education of radiology staff(10, 19, 27, 28). CONCLUSION In conclusion, when CTDIvol exceeded ~20 mGy for infants (<1 year) and 28 mGy for children from 1 to 9 years of age, 100% of CT images were deemed of diagnostic quality for routine head CT exams performed with multidetector systems. We believe these CTDIvol values represent a good balance between image quality and radiation burden. At our institution, we have adopted a standardized approach to pediatric head CT imaging based on the results of this study. Additionally, continued efforts are underway to optimize head CT imaging in pediatric patients. Our results pertain to head CT protocols with filtered back projection algorithm, and further studies should evaluate the relationship between CTDIvol and image quality for head CT protocols that employ iterative reconstruction and dose modulation techniques. IRB STATEMENT An appropriate institutional review board approved the study. GRANT SUPPORT None. PRESENTATION AT MEETINGS ASNR Meeting 2016, ‘Optimizing the Balance between Radiation Dose and Image Quality in Pediatric Head CT’. Abstract No: O-74. REFERENCES 1 Broder , J. , Fordham , L. A. and Warshauer , D. M. Increasing utilization of computed tomography in the pediatric emergency department, 2000–2006 . Emerg. Radiol. 14 ( 4 ), 227 – 232 ( 2007 ). Google Scholar CrossRef Search ADS PubMed 2 Markel , T. A. , Kumar , R. , Koontz , N. A. , Scherer , L. R. and Applegate , K. E. The utility of computed tomography as a screening tool for the evaluation of pediatric blunt chest trauma . J. Trauma 67 ( 1 ), 23 – 28 ( 2009 ). Google Scholar CrossRef Search ADS PubMed 3 Blackwell , C. D. , Gorelick , M. , Holmes , J. F. , Bandyopadhyay , S. and Kuppermann , N. Pediatric head trauma: changes in use of computed tomography in emergency departments in the United States over time . Ann. Emerg. Med. 49 ( 3 ), 320 – 324 ( 2007 ). Google Scholar CrossRef Search ADS PubMed 4 Brenner , D. J. and Hall , E. J. Computed tomography—an increasing source of radiation exposure . N. Engl. J. Med. 357 ( 22 ), 2277 – 2284 ( 2007 ). Google Scholar CrossRef Search ADS PubMed 5 White , K. S. Reduced need for sedation in patients undergoing helical CT of the chest and abdomen . Pediatr. Radiol. 25 ( 5 ), 344 – 346 ( 1995 ). Google Scholar CrossRef Search ADS PubMed 6 Shahi , V. , Brinjikji , W. , Cloft , H. J. , Thomas , K. B. and Kallmes , D. F. Trends in CT utilization for pediatric fall patients in US Emergency Departments . Acad. Radiol. 22 ( 7 ), 898 – 903 ( 2015 ). Google Scholar CrossRef Search ADS PubMed 7 Brenner , D. , Elliston , C. , Hall , E. and Berdon , W. Estimated risks of radiation-induced fatal cancer from pediatric CT . AJR Am. J. Roentgenol. 176 ( 2 ), 289 – 296 ( 2001 ). Google Scholar CrossRef Search ADS PubMed 8 United Nations Scientific Committee on the Effects of Atomic Radiation. Sources, Effects and Risks of Ionizing Radiation. I ( 2013 ). http://www.unscear.org/docs/reports/2013/13-85418_Report_2013_Annex_A.pdf 9 Brenner , D. J. Estimating cancer risks from pediatric CT: going from the qualitative to the quantitative . Pediatr. Radiol. 32 ( 4 ), 228 – 231 ( 2002 ) ; discussion 242–4. Google Scholar CrossRef Search ADS PubMed 10 Miglioretti , D. L. et al. . The use of computed tomography in pediatrics and the associated radiation exposure and estimated cancer risk . JAMA Pediatr. 167 ( 8 ), 700 – 707 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 11 Pearce , M. S. et al. . Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retrospective cohort study . Lancet 380 ( 9840 ), 499 – 505 ( 2012 ). Google Scholar CrossRef Search ADS PubMed 12 Mettler , F. A. , Jr. , Wiest , P. W. , Locken , J. A. and Kelsey , C. A. CT scanning: patterns of use and dose . J. Radiol. Prot. 20 ( 4 ), 353 – 359 ( 2000 ). Google Scholar CrossRef Search ADS PubMed 13 Yamauchi-Kawaura , C. , Fujii , K. , Aoyama , T. , Koyama , S. and Yamauchi , M. Radiation dose evaluation in head and neck MDCT examinations with a 6-year-old child anthropomorphic phantom . Pediatr. Radiol. 40 ( 7 ), 1206 – 1214 ( 2010 ). Google Scholar CrossRef Search ADS PubMed 14 Zacharias , C. , Alessio , A. M. , Otto , R. K. , Iyer , R. S. , Philips , G. S. , Swanson , J. O. and Thapa , M. M. Pediatric CT: strategies to lower radiation dose . AJR Am. J. Roentgenol. 200 ( 5 ), 950 – 956 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 15 Berrington de Gonzalez , A. , Mahesh , M. , Kim , K. P. , Bhargavan , M. , Lewis , R. , Mettler , F. and Land , C. Projected cancer risks from computed tomographic scans performed in the United States in 2007 . Arch. Intern. Med. 169 ( 22 ), 2071 – 2077 ( 2009 ). Google Scholar CrossRef Search ADS PubMed 16 American College of Radiology. Accreditation Program Requirements ( 2016 ). Available on http://www.acraccreditation.org/~/media/ACRAccreditation/Documents/CT/Requirements.pdf?la=en. 17 The Alliance for Radiation Safety in Pediatric Imaging. Image Gently: Computed Tomography ( 2014 ). Available on http://www.imagegently.org/Procedures/Interventional-Radiology/Protocols. 18 Tipnis , S. , Thampy , R. , Rumboldt , Z. , Spampinato , M. , Matheus , G. and Huda , W. Radiation intensity (CTDIvol) and visibility of anatomical structures in head CT examinations . J. Appl. Clin. Med. Phys. 17 ( 1 ), 293 – 300 ( 2016 ). Google Scholar CrossRef Search ADS PubMed 19 Paolicchi , F. , Faggioni , L. , Bastiani , L. , Molinaro , S. , Puglioli , M. , Caramella , D. and Bartolozzi , C. Optimizing the balance between radiation dose and image quality in pediatric head CT: findings before and after intensive radiologic staff training . AJR Am. J. Roentgenol. 202 ( 6 ), 1309 – 1315 ( 2014 ). Google Scholar CrossRef Search ADS PubMed 20 Linton , O. W. and Mettler , F. A. , Jr. National conference on dose reduction in CT, with an emphasis on pediatric patients . AJR Am. J. Roentgenol. 181 ( 2 ), 321 – 329 ( 2003 ). Google Scholar CrossRef Search ADS PubMed 21 Lee , C. , Pearce , M. S. , Salotti , J. A. , Harbron , R. W. , Little , M. P. , McHugh , K. , Chapple , C. L. and Berrington de Gonzalez , A. Reduction in radiation doses from paediatric CT scans in Great Britain . Br. J. Radiol. 89 ( 1060 ), 20150305 ( 2016 ). Google Scholar CrossRef Search ADS PubMed 22 McCollough , C. H. , Primak , A. N. , Braun , N. , Kofler , J. , Yu , L. and Christner , J. Strategies for reducing radiation dose in CT . Radiol. Clin. North Am. 47 ( 1 ), 27 – 40 ( 2009 ). Google Scholar CrossRef Search ADS PubMed 23 Kanal , K. M. , Graves , J. M. , Vavilala , M. S. , Applegate , K. E. , Jarvik , J. G. and Rivara , F. P. Variation in CT pediatric head examination radiation dose: results from a national survey . AJR Am. J. Roentgenol. 204 ( 3 ), W293 – W301 ( 2015 ). Google Scholar CrossRef Search ADS PubMed 24 Mullins , M. E. , Lev , M. H. , Bove , P. , O’Reilly , C. E. , Saini , S. , Rhea , J. T. , Thrall , J. H. , Hunter , G. J. , Hamberg , L. M. and Gonzalez , R. G. Comparison of image quality between conventional and low-dose nonenhanced head CT . AJNR Am. J. Neuroradiol. 25 ( 4 ), 533 – 538 ( 2004 ). Google Scholar PubMed 25 Pomerantz , S. R. , Kamalian , S. , Zhang , D. , Gupta , R. , Rapalino , O. , Sahani , D. V. and Lev , M. H. Virtual monochromatic reconstruction of dual-energy unenhanced head CT at 65-75 keV maximizes image quality compared with conventional polychromatic CT . Radiology 266 ( 1 ), 318 – 325 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 26 Kilic , K. , Erbas , G. , Guryildirim , M. , Konus , O. L. , Arac , M. , Ilgit , E. and Isik , S. Quantitative and qualitative comparison of standard-dose and low-dose pediatric head computed tomography: a retrospective study assessing the effect of adaptive statistical iterative reconstruction . J. Comput. Assist. Tomogr. 37 ( 3 ), 377 – 381 ( 2013 ). Google Scholar CrossRef Search ADS PubMed 27 Goske , M. J. et al. . The Image Gently campaign: working together to change practice . AJR Am. J. Roentgenol. 190 ( 2 ), 273 – 274 ( 2008 ). Google Scholar CrossRef Search ADS PubMed 28 De Bondt , T. , Mulkens , T. , Zanca , F. , Pyfferoen , L. , Casselman , J. W. and Parizel , P. M. Benchmarking pediatric cranial CT protocols using a dose tracking software system: a multicenter study . Eur. Radiol. 27 ( 2 ), 841 – 850 ( 2017 ). Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Radiation Protection DosimetryOxford University Press

Published: May 4, 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