Comparison of Radiation Dose and Image Quality of Abdominopelvic CT Using Iterative (AIDR 3D) and Conventional Reconstructions

Transcription

1 Medical Physics and Informatics Original Research Medical Physics and Informatics Original Research Caroline Duarte de Mello-Amoedo 1 Aparecido Nakano Martins 1 Adriano Tachibana 1 Daniella Ferraro Pinho 2 Ronaldo Hueb Baroni 1 Mello-Amoedo CD, Martins AN, Tachibana A, Pinho DF, Baroni RH Keywords: adaptive iterative dose reduction 3D, CT, iterative reconstruction, radiation dose reduction doi.org/ /ajr Received January 31, 2017; accepted after revision August 4, Radiology and Diagnostic Imaging Department, Hospital Israelita Albert Einstein, 627 Albert Einstein Ave, São Paulo, SP , Brasil. Address correspondence to C. D. Mello-Amoedo (amoedocaroline@gmail.com). 2 Department of Radiology, University of Texas Southwestern Medical Center, Dallas, TX. AJR 2018; 210: X/18/ American Roentgen Ray Society Comparison of Radiation Dose and Image Quality of Abdominopelvic CT Using Iterative (AIDR 3D) and Conventional Reconstructions OBJECTIVE. The purpose of this study is to compare radiation dose and image quality of abdominopelvic CT studies reconstructed with iterative and conventional techniques. MATERIALS AND METHODS. This retrospective study enrolled 99 patients who underwent abdominopelvic CT examinations with the portal venous phase images reconstructed with both filtered back projection and Adaptive Iterative Dose Reduction 3D (AIDR 3D) at different time points. Subjective assessment of image quality was performed by two radiologists who scored axial images for overall quality, sharpness, noise, and acceptability in a blinded fashion. The SD of the mean attenuation of the liver, aorta, and paraspinal muscle (as a measurement of image noise) and contrast-to-noise and signal-to-noise ratios for liver and aorta were used as objective parameters of image quality. Radiation dose parameters included CT dose index volume (CTDI vol ), dose-length product, effective dose (ED), and size-specific dose estimate (SSDE). Results were compared for different body mass index (BMI; weight in kilograms divided by the square of height in meters) categories. Paired t test and McNemar paired tests for noninferiority were used, with p < 0.05 considered statistically significant. RESULTS. We obtained a 62.5% mean reduction in CTDI vol, a 58% mean reduction in ED, and a 63% mean reduction in SSDE when AIDR 3D was used (p < 0.001). Subjective parameters of image quality were considered noninferior for AIDR 3D studies compared with filtered back projection (p < 0.001), except for the sharpness of images of patients with BMI Variable results were found regarding objective assessment of image quality. CONCLUSION. AIDR 3D allowed a significant reduction in radiation dose of abdominopelvic CT examinations without a loss of image quality in general. C T has revolutionized medical imaging. Since its first use in the 1970s, countless technologic advances have been implemented and an impressive expansion of the clinical applications of this method has occurred, so that it now affects all levels of patient care. As key breakthroughs, we highlight improvements in CT hardware capacity and data-processing methods, which were determinants for the faster acquisition times, better scanning plane resolution, and shorter image reconstruction times achieved by the successive generations of CT scanners. The first CT scanner was used only for head studies and offered low spatial resolution of images, limiting the number of abnormalities that could be analyzed. The most modern scanners (sixth generation) have helical systems and multiple detectors that are capable of scanning any desired segment of the body in few seconds and generating images with isotro- pic spatial resolution (thickness of 1 mm) that allow high-quality multiplanar and 3D reconstructions, which are fundamental features for the utilization of CT in situations not previously imagined (e.g., cardiac and vascular diseases) [1]. Thus, a rapid and growing expansion in the number of scans performed has occurred. In 2014, it was estimated that approximately 81 million CT scans were performed in the United States, compared with 62 million in 2006 and 3 million in 1980 [1, 2]. These numbers have raised concerns about the cumulative radiation dose associated with CT examinations because of a potentially increased risk of induction of cancer, especially in highly vulnerable patients, such as the pediatric population. Although many alarmist articles regarding this topic have been published [3 6], even when large population cohorts are analyzed [7, 8], a critical review of the available data sup- AJR:210, January

2 ports that this risk, if it exists, is very low and probably outweighed by the benefits of CT if the examinations are properly indicated (i.e., the justification principle of radiation protection) [9, 10]. Thus, a reasonable approach assumes minimization of radiation to the lowest dose consistent with acquisition of the desired information as a top priority to fully use the benefits provided by state-of-the-art CT technology (i.e., optimization principle of radiation protection). This is now what many respectable imaging organizations (e.g., American Association of Physicists in Medicine, American College of Radiology, and Radiological Society of North America) recommend [11]. In this context, several strategies for radiation dose reduction in CT have already been successfully incorporated into clinical practice [12]. The most effective one, based on reduction of x-ray beam energy, is hindered by increased image noise and degraded image quality, mainly associated with the limitations of the standard filtered back projection (FBP) reconstruction algorithm that has been used in all devices since the development of CT [13]. As with all analytic reconstruction algorithms, the FBP is based on an exact mathematic relationship between the acquired data and the reconstructed image without statistical noise consideration. FBP also ignores the modeling processes (x-ray beam geometry and photon interactions as scanned object and receiver), assuming that the projection data are free of noise when, in fact, they are not. Therefore, FBP is usually acceptable only when standard radiation doses are used. When the radiation doses are substantially reduced and the photon intensity decreases, image noise is amplified. In addition, artifacts will emerge, along with limitations in spatial resolution, thus worsening image quality [14]. Iterative reconstruction algorithms have recently been integrated with the automatic exposure control systems of the most modern CT scanners from the major vendors, with the purpose of allowing low-dose examinations without a loss of image quality [15]. Considering that the doses delivered in abdominal scans are the highest ones compared with other regions of the body (e.g., the exposure from a head CT scan is approximately 1 2 msv, whereas the mean exposure from a single-phase body CT scan is approximately 14 msv, and this single-phase dose can add up to msv for abdominal examinations requiring multiple scans at different contrast enhancement phases) [16, 17] and that these scans comprise up to one-third of all CT examinations, these algorithms would be quite desirable; however, their costs are still limiting for various imaging services (around $100,000 for incorporating iterative reconstruction technique if the scanner is compatible, and around $1,000,000 to buy a scanner that already has it). Thus, studies focusing on radiation dosereduction rates obtained with these new algorithms and their implications for image quality in different clinical scenarios are needed, but are still relatively scarce in the literature [18, 19]. Our purpose was to compare radiation dose and image quality of abdominopelvic CT examinations obtained with an iterative reconstruction algorithm from a specific vendor (Adaptive Iterative Dose Reduction 3D [AIDR 3D], Toshiba Medical Systems) with images from the same patients reconstructed using FBP. Materials and Methods Study Design This was a single-center retrospective study approved by the ethics committee of Hospital Israelita Albert Einstein. Informed consent was waived. We performed a database search for patients who underwent contrast-enhanced abdominopelvic CT for any indication on a 320-MDCT scanner (Aquilion ONE, Toshiba Medical Systems) during the first 15 months after the introduction of the AIDR 3D algorithm (February 2012 to May 2013), who previously had undergone abdominopelvic CT performed on the same scanner using conventional reconstruction (FBP). Studies of patients younger than 18 years, examinations without portal venous phase images, interval greater than 24 months between examinations, or patients with a change in body mass index (BMI; weight in kilograms divided by the square of height in meters) category between AIDR 3D and FBP examinations were excluded from the study. Acquisition and Reconstruction Parameters All examinations were obtained with similar acquisition parameters, except for beam collimation (set at for AIDR 3D and for FBP studies) and noise level (set at 11.5 for AIDR 3D and 12.5 for FBP studies), which were set when AIDR 3D was installed according to the recommendations of the vendor (Table 1). The noise level is a parameter that guides the modulation of tube current performed by the automatic exposure control integrated with Aquilion ONE ( Sure Exposure 3D, Toshiba Medical Systems); as long as the CT gantry moves around the patient, the tube current will be delivered according to the thickness of the region of the patient, and the software will calculate a desired image quality on the basis of the level of noise setting. For both acquisitions, IV iodinated contrast agent (iobitridol 350 mg I/mL; Xenetix, Guerbet) was given at a dose of 1.3 ml/kg. The acquisition extended from the bases of the lungs to the pubic symphysis. The difference in extension of both studies was recorded to assess for eventual variance between AIDR 3D and FBP examinations. All AIDR 3D scans were set on standard mode at the time of imaging acquisition (referring to the level or strength of radiation dose reduction for this specific iterative reconstruction algorithm, also available in mild and strong modes), as recommended by the vendor. Subjective Assessment of Image Quality Two board-certified abdominal radiologists independently reviewed all axial slices of the 3-mm reconstructed portal venous phase images, randomly and blinded to the type of reconstruction (FBP or AIDR 3D). The studies were presented in the same order for both radiologists to prevent bias. CT images were anonymized for patient information, date of acquisition, and scanning parameters and were presented to the reviewers at a PACS station in one sitting. A 5-point grading scale was used to classify the examinations (based on all slices) in terms of overall quality (1, unacceptable; 2, poor; 3, moder- TABLE 1: Acquisition Parameters of the Abdominopelvic CT Studies Reconstructed with AIDR 3D FBP Acquisition Parameter AIDR 3D FBP Tube potential (kvp) Tube current (ma) AEC ( Sure Exposure 3D) AEC ( Sure Exposure 3D) Noise level Beam collimation (mm) Pitch Rotation time (s) Note Sure Exposure 3D and AIDR 3D are products of Toshiba Medical Systems. AIDR = Adaptive Iterative Dose Reduction, FBP = filtered back projection, AEC = automatic exposure control AJR:210, January 2018

3 Fig year-old man with epigastric pain. Axial abdominal CT image was obtained at level of celiac trunk. Image was reconstructed with Adaptive Iterative Dose Reduction 3D (Toshiba Medical Systems). ROIs (circles) were positioned in right lobe of liver (L), aorta (A), and right paravertebral muscle (M). ate; 4, good; 5, excellent), sharpness (1, unacceptable; 2, poor; 3, moderate; 4, good; 5, excellent), and noise (1, excessive; 2, major; 3, moderate; 4, minimal; 5, none). Images were also rated as acceptable with or without restrictions. A training session based on 10 abdominopelvic CT examinations was given to both radiologists with the intent of making them familiar with the scoring system before the image quality assessment started. Objective Assessment of Image Quality A third board-certified abdominal radiologist used a single axial section (at the level of the celiac trunk origin) of each CT examination to obtain the mean attenuation (in Hounsfield units) and the SD of the mean attenuation of the liver, aorta, and muscle (considered as the objective image noise). Circular ROIs were manually posi- A B Fig year-old man (body mass index [weight in kilograms divided by the square of height in meters], 25) under oncologic surveillance. A and B, Axial portal phase CT images were reconstructed with filtered back projection (A) and Adaptive Iterative Dose Reduction 3D (AIDR 3D; Toshiba Medical Systems) (B). In both panels, top image shows superior abdomen, and bottom image shows pelvis. Both reconstructions were rated as having good overall quality, good sharpness, and minimal noise and as being acceptable without restrictions by both reviewers, but AIDR 3D algorithm allowed 57% reduction in effective dose (from 17.3 to 7.5 msv) and 61% reduction in size-specific dose estimate (from 28.4 to 11.2 mgy) in this case. AJR:210, January

4 tioned as follows (Fig. 1): in the right hepatic lobe, with an area of 150 ± 20 mm 2, avoiding vessels and focal lesions (as a reference, we used a horizontal imaginary line drawn in the plane of the aorta, leaving the ROI in the more peripheral portion of the parenchyma); in the aorta, including at least two-thirds of the circumference of the vessel and avoiding its walls; and in the right paraspinal muscle, with an area of 150 ± 20 mm 2. The values obtained were then used to calculate the signalto-noise ratio (SNR) and contrast-to-noise ratio (CNR) of liver and aorta, according to the following equations: SNR = ROI o / SD o, and CNR = (ROI o ROI m ) / SD m, where ROI o is the mean attenuation of the organ of interest, SD o is the SD of the organ of interest, ROI m is the mean attenuation of paraspinal muscle, and SD m is the SD of the paraspinal muscle. Radiation Dose The volume CT dose index (CTDI vol ) and dose-length product (DLP) generated on dose reports by the scanner were recorded for each examination. The radiation effective dose (ED) was calculated by multiplying DLP by an abdominal-specific correction coefficient (0.015 msv/mgy cm) [20]. The anteroposterior and lateral abdominal diameters (in centimeters) of each patient in both examinations were measured by the same radiologist on a PACS station using the same axial sections (at the level of the celiac trunk origin) used to position the ROIs for the objective measurement of image noise. The effective diameters were then calculated to determine the 32-cm phantom conversion factor for obtaining the size-specific dose estimate (SSDE) from the CTDI vol [21]. Patients were divided into three BMI categories (< 20, , and 25) for comparison of radiation dose and image quality. Statistical Analysis Data were analyzed using R (version 3.2.2, R Foundation for Statistical Computing) and SPSS (version 17.0, IBM). Paired t tests were used to compare radiation dose measurements and objective image quality parameters. The Gwet coefficient was used to evaluate agreement between the two reviewers regarding subjective image quality [22]. McNemar paired tests were used to verify the noninferiority hypothesis of the subjective image quality of AIDR 3D examinations in comparison with FBP, considering as the null hypothesis one that shows that the image quality of AIDR 3D examinations is inferior to that of FBP examinations, and as an alternative hypothesis, one that shows that the image quality of AIDR 3D examinations is not inferior to that of FBP examinations [23]. We adopted a noninferiority margin of 15% and the results obtained by the reviewer with greater professional experience. A 5% level of significance was considered. Results A total of 198 abdominopelvic CT examinations (99 reconstructed with AIDR 3D and 99 reconstructed with FBP) from 99 patients (52 men and 47 women) were included. The mean age of the patients was 57.5 ± 14.6 years (range, years) at the time of AIDR 3D examination and 56.7 ± 14.5 years (range, years) at the time of FBP examination. The median BMI was 26 (range, 18 40). Patients were divided into three BMI categories: less than 20 (n = 5), (n = 36), and greater than or equal to 25 (n = 58). The median interval between the two scans was 11.7 months ( months). There was no significant difference in acquisition lengths between AIDR 3D and FBP CT examinations (44.5 ± 3.1 cm and 44.5 ± 3.3 cm, respectively; p = 0.611). Radiation Dose All radiation dose parameters were significantly lower for AIDR 3D compared with FBP scans, for the entire population and also when the patients were divided according to BMI category (p < 0.001) (Table 2). The mean reduction in CTDI vol was 62.5% for patients with BMI less than 20, 74% for patients with BMI , 72% for patients with BMI greater than or equal to 25, and 56% for all patients. For DLP and ED, mean reductions were 58% for all patients, 63% for patients with BMI less than 20, 67% for patients with BMI , and 52% for patients with BMI greater than or equal to 25. For SSDE, mean reductions were 63% for all patients, 74% for patients with BMI less than 20, 72% for patients with BMI , and 57% for patients with BMI greater than or equal to 25. Subjective Assessment of Image Quality All examinations were classified by both reviewers with scores equal to or greater than 3 for overall quality (moderate, good, or excellent), sharpness (moderate, good, or excellent), and noise (moderate, minimal, or none). For an example, see Figure 2. Thus, we grouped scores 1 and 2 and scores 4 and 5 for statistical analysis. Interobserver concordance was good to excellent (Gwet coefficient, ). In the total population, we showed the noninferiority of the AIDR 3D technique regarding all subjective parameters of image quality (overall quality, sharpness, noise, and acceptability), with p < When the BMI categories were analyzed separately, we observed evidence of the noninferiority of AIDR 3D examinations for all subjective image quality parameters except for sharpness in the BMI category. It was not possible to apply noninferiority testing in the BMI less than 20 category because of the small number of patients in this group (n = 5). These results are presented in Table 3. Objective Assessment of Image Quality Variable results were obtained regarding objective parameters of image quality. For TABLE 2: Radiation Dose Parameters of Abdominopelvic CT Examinations Reconstructed With AIDR 3D and FBP Radiation Dose All Patients (n = 99) BMI < 20 (n = 5) BMI (n = 36) BMI 25 (n = 58) Parameter AIDR 3D FBP AIDR 3D FBP AIDR 3D FBP AIDR 3D FBP CTDI vol (mgy) 9.9 ± ± ± ± ± ± ± ± 7.7 DLP (mgy cm) ± ± ± ± ± ± ± ± ED (msv) 7.5 ± ± ± ± ± ± ± ± 6.2 SSDE (mgy) 11.9 ± ± ± ± ± ± ± ± 6.9 Note Data are mean ± SD. For all dose parameters, the differences between AIDR 3D and FBP reconstructed examinations were statistically significant (p < 0.001). AIDR = Adaptive Iterative Dose Reduction (Toshiba Medical Systems), FBP = filtered back projection, BMI = body mass index (weight in kilograms divided by the square of height in meters), CTDI vol = CT dose index volume, DLP = dose-length product, ED = effective dose, SSDE = size-specific dose estimate. 130 AJR:210, January 2018

5 TABLE 3: Subjective Assessment of Image Quality of Abdominopelvic CT Examinations Reconstructed With AIDR 3D and FBP All Patients (n = 99) BMI < 20 (n = 5) a BMI (n = 36) BMI 25 (n = 58) Subjective Parameters all patients, AIDR 3D had lower noise and greater SNR compared with FBP examinations considering only the aorta. Apart from that, AIDR 3D examinations showed lower CNR for the liver compared with FBP examinations. These results and those for the three BMI categories are shown in Table 4. Discussion Our results showed a mean substantial reduction in radiation dose (62.5% for CTDI vol, AIDR 3D FBP p AIDR 3D FBP AIDR 3D FBP p AIDR 3D FBP p Overall quality < < < Excellent or good 98 (99) 97 (98) 5 (100) 5 (100) 36 (100) 35 (97) 57 (98) 57 (98) Moderate 1 (1) 2 (2) 0 (0) 0 (0) 0 (0) 1 (3) 1 (2) 1 (2) Poor or unacceptable 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Sharpness < > 0.99 < Excellent or good 94 (95) 96 (97) 5 (100) 4 (80) 32 (89) 35 (97) 57 (98) 57 (98) Moderate 5 (5) 3 (3) 0 (0) 1 (20) 4 (11) 1 (3) 1 (2) 1 (2) Poor or unacceptable 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Noise < < < None or minimal 95 (96) 83 (84) 4 (80) 2 (40) 35 (97) 33 (92) 56 (97) 48 (83) Moderate 4 (4) 16 (16) 1 (20) 3 (60) 1 (3) 3 (8) 2 (3) 10 (17) Major or excessive 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) Acceptability < < < Without restrictions 96 (97) 97 (98) 5 (100) 5 (100) 34 (94) 35 (97) 57 (98) 57 (98) With restrictions 3 (3) 2 (2) 0 (0) 0 (0) 2 (6) 1 (3) 1 (2) 1 (2) Note Data are number (percentage) of examinations. AIDR = Adaptive Iterative Dose Reduction (Toshiba Medical Systems), FBP = filtered back projection, BMI = body mass index (weight in kilograms divided by the square of height in meters). a Noninferiority testing was not applicable to this BMI category. 58% for ED, and 63% for SSDE) for abdominopelvic CT examinations reconstructed with AIDR 3D compared with those using FBP, without a loss of subjective image quality (all parameters evaluated were not considered inferior). These observations were similar to those of other authors who also studied the AIDR 3D technique in abdominal scans. Matsuki et al. [19] prospectively evaluated 60 patients who underwent upper abdominal CT for hepatic metastases screening and found a mean reduction in radiation dose of 49.2%, without showing a significant difference in the subjective evaluation of the image quality when comparing AIDR 3D and FBP reconstructions. Gervaise et al. [18], in a retrospective study of 21 patients who underwent abdominopelvic CT scans, showed a mean reduction in radiation dose of 49.5% when the AIDR 3D technique was used, also without showing a significant difference between the mean scores obtained on the subjective TABLE 4: Objective Assessment of Image Quality of Abdominopelvic CT Examinations Reconstructed With AIDR 3D and FBP Objective Parameter, Organ All Patients (n = 99) BMI < 20 (n = 5) BMI (n = 36) BMI 25 (n = 58) AIDR 3D FBP p AIDR 3D FBP p AIDR 3D FBP p AIDR 3D FBP p Noise Liver 12.9 ± ± ± ± ± ± 1.5 < ± ± Aorta 14.9 ± ± 3.3 < ± ± ± ± ± ± 3.7 < Muscle 13.7 ± ± ± ± ± ± 1.5 < ± ± SNR Liver 8.0 ± ± ± ± ± ± ± ± Aorta 9.5 ± ± ± ± ± ± ± ± 1.7 < CNR Liver 3.1 ± ± ± ± ± ± 1.3 < ± ± Aorta 5.8 ± ± ± ± ± ± ± ± Note Data are mean ± SD. AIDR = Adaptive Iterative Dose Reduction (Toshiba Medical Systems), FBP = filtered back projection, BMI = body mass index (weight in kilograms divided by the square of height in meters), SNR = signal-to-noise ratio, CNR = contrast-to-noise ratio. AJR:210, January

6 evaluation of image quality. It is important to mention that the higher rates of dose reduction achieved in our study in comparison with the others we have mentioned should reflect differences in conventional (i.e., FBP) CT doses (our examinations provided higher radiation doses, and thus even a similar dose reduction might appear to be greater) and acquisition protocol (e.g., our noise parameter was set at a higher level for AIDR 3D examinations, which means that we tolerated more noise and therefore allowed higher radiation dose reduction). In this context, it is reasonable to discuss the possible effects of the different beam collimation and noise level settings of AIDR 3D and FBP examinations in our study to make it clear that no bias was introduced in our data: for a narrower beam collimation (as used in AIDR 3D examinations), higher radiation doses would be expected [13], so although beam collimation was not equal among conventional and reduced-dose examinations, this parameter might not have favored the iterative reconstruction technique we tested. The same notion is valid for the noise parameter, because even when we used a lower noise level ( ), AIDR 3D provided significantly lower radiation doses [24]. Because of this technical issue, it is difficult to compare our results with those obtained using other iterative reconstruction techniques, but they are all encouraging: Sagara et al. [25] and Prakash et al. [26] showed that it was possible to reduce abdominal CT scan doses by 33% and 25%, respectively, using adaptive statistical iterative reconstruction (AISR), while improving image quality. May et al. [27] found a 50% reduction in abdominal scan dose using iterative reconstruction in image space compared with standard FBP reconstructions, with equivalent image quality. Regarding image quality, it is also important to mention that none of the available studies in the literature used noninferiority tests to compare 3D AIDR and FBP techniques, concluding that there was similarity between them in terms of subjective image quality when p > 0.05, which we think is questionable. When the patients were grouped according to BMI, our results corroborated the great variability in terms of radiation dose reduction and evaluation of subjective image quality observed in the literature. Matsuki et al. [19] found the highest and lowest dose-reduction rates in AIDR 3D examinations performed for patients with BMI greater than or equal to 25 and patients with BMI of , respectively. On the other hand, our results were similar to those achieved by Juri et al. [28], who found that, for nephrographic phase images of CT urography examinations reconstructed with AIDR 3D, the largest and smallest reductions in CTDI vol occurred when patients had a BMI of and greater than or equal to 25, respectively. However, in that same study, the result was the opposite when the radiation doses of the excretory phase were considered. Conflicting results regarding dose reduction in relation to BMI also appeared in studies evaluating iterative reconstruction techniques from other manufacturers [25]. In our study, the image sharpness of examinations reconstructed with the AIDR 3D technique was considered inferior to that of FBP-reconstructed images when patients had a BMI of ; however, this finding was not accompanied by a loss of general image quality or acceptability. Although it is not related to a specific category of BMI, the worst sharpness of AIDR 3D examinations compared to those reconstructed with FBP has already been found by other authors who evaluated CT urography [28] and CT enterography [29] examinations. As it was considered in those studies, the smoothing appearance of AIDR 3D images could lead to some difficulty in visualizing fine details. Regarding the objective evaluation of image quality, we obtained lower noise for AIDR 3D examinations considering only the aorta, which differs partially from the literature. Matsuki et al. [19] found lower mean values of noise for AIDR 3D examinations of all organs evaluated (aorta, portal vein, liver, and pancreas), whereas Gervaise et al. [18] found no difference when they considered the aorta and the liver. By BMI category, the results were also conflicting. In practice, however, what would matter most would be the subjective perception of noise, and this was not considered worse for AIDR 3D examinations. Perhaps this divergence between objective and subjective measures of noise can be explained by the nonstandardization of target locations for subjective analysis, just as it was done for objective noise analysis. These results may be better evaluated in future studies. There are limitations to our study. First, its design is retrospective, which means that we compared examinations previously performed at different dates and for which we had no control over the acquisition parameters. We believe, however, that prospective studies would not be easily justified in the scenario of radiation dose reduction for CT. We also did not include pediatric patients, a target population for radiation dose-reduction strategies. Considering all the peculiarities of tomographic acquisition parameters for pediatric patients, it seems reasonable that this group should be evaluated separately, which would have limited the number of patients in our study. We did not evaluate the diagnostic accuracy of the CT examinations, which would be a more appropriate approach in specific clinical contexts (e.g., detection of metastases in patients with cancer). Finally, we were not able to test the noninferiority of the subjective image quality of the examinations performed of patients with BMI less than 20, because this category was not adequately represented. In conclusion, abdominopelvic CT examinations using AIDR 3D in our study population enabled substantial reduction of radiation dose compared with routine-dose CT using FBP, in general, without a loss of image quality. Acknowledgments We thank Ana Carolina Cintra Nunes Mafra and Elivane da Silva Victor for helping with the statistical analyses. References 1. Mahesh M. Search for isotropic resolution in CT from conventional through multiple-row detector. RadioGraphics 2002; 22: Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med 2007; 357: Fazel R, Krumholz HM, Wang Y, et al. Exposure to low-dose ionizing radiation from medical imaging procedures. N Engl J Med 2009; 361: Sodickson A, Baeyens PF, Andriole KP, et al. Recurrent CT, cumulative radiation exposure, and associated radiation-induced cancer risks from CT of adults. Radiology 2009; 251: Smith-Bindman R, Lipson J, Marcus R, et al. Radiation dose associated with common computed tomography examinations and the associated lifetime attributable risk of cancer. Arch Intern Med 2009; 169: Berrington de González A, Mahesh M, Kim KP, et al. Projected cancer risks from computed tomographic scans performed in the United States in Arch Intern Med 2009; 169: Mathews JD, Forsythe AV, Brady Z, et al. Cancer risk in 680,000 people exposed to computed tomography scans in childhood or adolescence: data linkage study of 11 million Australians. BMJ 2013; 346:f Pearce MS, Salotti JA, Little MP, et al. Radiation exposure from CT scans in childhood and subsequent risk of leukaemia and brain tumours: a retro- 132 AJR:210, January 2018

7 spective cohort study. Lancet 2012; 380: Gervaise A, Osemont B, Louis M, Lecocq S, Fitzgerald NE, Kundra V. Quality initiatives: CT 9. McCollough CH, Guimaraes L, Fletcher JG. In Teixeira P, Blum A. Standard dose versus low-dose radiation dose reduction how to implement defense of body CT. AJR 2009; 193:28 39 abdominal and pelvic CT: comparison between fil- change without sacrificing diagnostic quality. 10. McCollough CH, Bushberg JT, Fletcher JG, Eckel tered back projection versus adaptive iterative dose RadioGraphics 2011; 31: LJ. Answers to common questions about the use reduction 3D. Diagn Interv Imaging 2014; 95: Sagara Y, Hara AK, Pavlicek W, Silva AC, Paden and safety of CT scans. Mayo Clin Proc 2015; 90: Verdun FR, Bochud F, Gundinchet F, Aroua A, Schnyder P, Meuli R. Quality initiatives radiation risk: what you should know to tell your patient. RadioGraphics 2008; 28: Lee TY, Chhem RK. Impact of new technologies on dose reduction in CT. Eur J Radiol 2010; 76: McNitt-Gray MF. AAPM/RSNA physics tutorial for residents: topics in CT radiation dose in CT. RadioGraphics 2002; 22: Fleischmann D, Boas FE. Computed tomography: old ideas and new technology. Eur Radiol 2011; 21: Jensen K, Martinsen AC, Tingberg A, Aalokken TM, Fosse E. Comparing five different iterative reconstruction algorithms for computed tomography in an ROC study. Eur Radiol 2014; 24: Mettler FA Jr, Huda W, Yoshizumi TT, Mahesh M. Effective doses in radiology and diagnostic nuclear medicine: a catalog. Radiology 2008; 248: Costello JE, Cecava ND, Tucker JE, Bau JL. CT radiation dose: current controversies and dose reduction strategies. AJR 2013; 201: Matsuki M, Murakami T, Juri H, Yoshikawa S, Narumi Y. Impact of adaptive iterative dose reduction (AIDR) 3D on low-dose abdominal CT: comparison with routine-dose CT using filtered back projection. Acta Radiol 2013; 54: Jessen K, Panzer W, Shrimpton P, et al. EUR 16262: European guidelines on quality criteria for computed tomography. Luxembourg City, Luxembourg: Office for Official Publications of the European Communities, Boone JM, Strauss KJ, Cody DD, et al. Size-specific dose estimates (SSDE) in pediatric and adult body CT examinations: report of AAPM Task Group 204. American Association of Physicists in Medicine website. Published Accessed August 24, Wongpakaran N, Wongpakaran T, Wedding D, Gwet KL. A comparison of Cohen s Kappa and Gwet s AC1 when calculating inter-rater reliability coefficients: a study conducted with personality disorder samples. BMC Med Res Methodol 2013; 13: Ahn S, Park SH, Lee KH. How to demonstrate similarity by using noninferiority and equivalence statistical testing in radiology research. Radiology 2013; 267: Tamm EP, Rong XJ, Cody DD, Ernst RD, RG, Wu Q. Abdominal CT: comparison of lowdose CT with adaptive statistical iterative reconstruction and routine-dose CT with filtered back projection in 53 patients. AJR 2010; 195: Prakash P, Kalra MK, Kambadakone AK, et al. Reducing abdominal CT radiation dose with adaptive statistical iterative reconstruction technique. Invest Radiol 2010; 45: May MS, Wust W, Brand M, et al. Dose reduction in abdominal computed tomography: intraindividual comparison of image quality of full-dose standard and half-dose iterative reconstructions with dual-source computed tomography. Invest Radiol 2011; 46: Juri H, Matsuki M, Inada Y, et al. Low-dose computed tomographic urography using adaptive iterative dose reduction 3-dimensional: comparison with routine-dose computed tomography with filtered back projection. J Comput Assist Tomogr 2013; 37: Wallihan DB, Podberesky DJ, Sullivan J, et al. Diagnostic performance and dose comparison of filtered back projection and adaptive iterative dose reduction three-dimensional CT enterography in children and young adults. Radiology 2015; 276: AJR:210, January