Medical Physics and Informatics Original Research

Medical Physics and Informatics Original Research Sagara et al. Abdominal CT With Low Versus Routine Dose Medical Physics and Informatics Original Research Yoshiko Sagara 1 Amy K. Hara 2 William Pavlicek 2 Alvin C. Silva 2 Robert G. Paden 2 Qing Wu 3 Sagara Y, Hara AK, Pavlicek W, Silva AC, Paden RG, Wu Q Keywords: CT, iterative reconstruction, radiation dose DOI:10.2214/AJR.09.2989 Received April 30, 2009; accepted after revision February 5, 2010. A. K. Hara, W. Pavlicek, and A. C. Silva have a research agreement with GE Healthcare. 1 Department of Radiology, Oita University Faculty of Medicine, Idaigaoka Hasama-machi, Yufu-shi, Oita, Japan. 2 Department of Radiology, Mayo Clinic, 13400 E Shea Blvd., Scottsdale, AZ 85259. Address correspondence to A. K. Hara (hara.amy@mayo.edu). 3 Department of Biostatistics, Mayo Clinic, Scottsdale, AZ. AJR 2010; 195:713 719 0361 803X/10/1953 713 American Roentgen Ray Society Abdominal CT: Comparison of Low-Dose CT With Adaptive Statistical Iterative Reconstruction and Routine-Dose CT With Filtered Back Projection in 53 Patients OBJECTIVE. The purpose of this article is to retrospectively compare radiation dose, noise, and image quality of abdominal low-dose CT reconstructed with adaptive statistical iterative reconstruction (ASIR) and routine-dose CT reconstructed with filtered back projection (FBP). MATERIALS AND METHODS. Fifty-three patients (37 men and 16 women; mean age, 60.8 years) underwent contrast-enhanced abdominal low-dose CT with 40% ASIR. All 53 patients had previously undergone contrast-enhanced routine-dose CT with FBP. With the scanning techniques masked, two radiologists independently graded images for sharpness, image noise, diagnostic acceptability, and artifacts. Quantitative measures of radiation dose and image noise were also obtained. All results were compared on the basis of body mass index (BMI). RESULTS. The volume CT dose index (CTDI vol ), dose length product, and radiation dose for low-dose CT with ASIR were 17 mgy, 860 mgy, and 13 msv, respectively, compared with 25 mgy, 1,193 mgy, and 18 msv for routine-dose CT with FBP, representing an approximate overall dose reduction of 33%. Low-dose CT with ASIR had significantly reduced (p < 0.001) quantitative and qualitative assessment of image noise. Image sharpness, however, was significantly reduced for low-dose CT with ASIR (p < 0.001), although diagnostic acceptability and artifact scores were nearly identical to those for routine-dose CT with FBP. The average CTDI vol dose reduction was 66% for patients with a BMI of less than 20 and 23% for patients with a BMI of 25 or greater. CONCLUSION. Compared with routine-dose CT with FBP, abdominal low-dose CT with ASIR significantly reduces noise, thereby permitting diagnostic abdominal examinations with lower (by 23 66%) radiation doses. Despite reduced image sharpness in average and small patients, low-dose CT with ASIR had diagnostic acceptability comparable to that of routine-dose CT with FBP. D uring 2007, more than 68 million CT scans were obtained in the United States (up from 3 million in 1980) [1, 2]. This increase in CT utilization has raised concern about the increasing risk of cancer from medical radiation exposures because organ doses with CT are often higher than those with other imaging tests [3]. For example, the radiation dose from an abdominal CT scan is at least 50 times greater than that from an abdominal radiograph [3]. Dose reduction techniques, such as tube current modulation, low tube voltage, and noise reduction filters, have been successfully implemented and have been shown to reduce radiation exposure [4 12]. However, further reductions in radiation dose are hindered by increased image noise and degraded image quality mainly as a result of limitations of the standard filtered back projection (FBP) reconstruction algorithm currently used on all CT scanners. Adaptive statistical iterative reconstruction (ASIR) is a newly introduced reconstruction algorithm for CT that, unlike FBP, uses a statistical model to reduce image noise [13]. A pilot study showed that, for low-dose CT scans, image quality was superior with ASIR than with FBP [14]. It is less clear whether low-dose CT with ASIR is comparable to routine-dose CT with FBP, although the initial results in 12 patients are encouraging [14]. The ability to acquire high-quality low-dose contrast-enhanced CT scans that benefit from the noise-lowering ability of ASIR could allow nearly all CT protocols to be prescribed using a reduced radiation dose in comparison with current techniques. The purpose of this study was to compare image quality of two different abdominal CT protocols in the same AJR:195, September 2010 713

Sagara et al. patients: low-dose contrast-enhanced CT reconstructed with ASIR and routine-dose contrast-enhanced CT reconstructed with FBP. Materials and Methods Study Design This retrospective study was compliant with the HIPAA and was approved by the institutional review board of the institution where the patients were examined. Patient Population Between December 2008 and January 2009, 169 contrast-enhanced venous phase abdominal lowdose CT with ASIR examinations were performed. Of these 169 examinations, 53 patients (37 men and 16 women; average age, 60.8 years; range, 20 86 years) had previously undergone contrastenhanced venous phase abdominal routine-dose CT with FBP. These 53 patients who underwent both low-dose CT with ASIR and routine-dose CT with FBP served as the study group. Low-Dose CT With ASIR Scanning Technique All low-dose CT scans were performed on a 64-MDCT scanner (Discovery CT750HD, GE Healthcare) using the following parameters: fixed noise index of 30.9; 0.625-mm collimation; reconstruction slice thickness of 5 mm (n = 16) or 3.75 mm (n = 37); 120 kvp; variable milliamperage determined by x-, y-, and z-axis dose modulation; gantry rotation time of 0.5 seconds; and 40% ASIR. All CT scans imaged the abdomen or both the abdomen and the pelvis, were obtained with IV contrast material, and were conducted during the venous phase (70 seconds). Rationale for Increased Noise Index As radiation dose decreases by 1/x, the image noise increases by the square root of x [15]. Therefore, when reducing the radiation dose by one half, the noise should increase by the square root Noise (SD) 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.1 2 4 Spatial Frequency (line pairs/cm) 6 of 2 (i.e., 1.4). We used this formula to calculate a probable noise index, which would result in a dose reduction of approximately 50%. Our standard noise index for 0.625-mm slices is 22.1. Multiplying 22.1 by 1.4 results in a noise index of 30.9, which, theoretically, should achieve an approximate 50% dose reduction in a patient of average size. Using a higher noise index is necessary for low-dose imaging because it causes the scanner to automatically select a lower milliamperage setting (and thus a lower dose) using dose-modulation software available on the scanner. This enables an automatic dose correction for patient size and is the same method used in the previous examinations with FBP. We adjusted the milliamperage range (a part of the noise index tool specification) to a minimum of 100 mas and a maximum of 650 mas to enable the prescription of noise index for patients of different sizes. The noise index is unique to slice thickness. In general, a higher noise index is used for thinner slices and a lower noise index is used for thicker slices. This is because the thin (< 1 mm) slices are typically not used for diagnostic purposes but used for multiplanar reconstructions. When the noisy thin slices are reconstructed to 3 mm, the image noise decreases and is typically diagnostically acceptable. In our practice, we choose the noise index according to a slice thickness of 0.625 mm, which is used for axial, coronal, and sagittal reconstructions at 3 mm. If only axial thicknesses of 3.75 mm and 5 mm are reconstructed, these could have lower noise index settings. Rationale for Percentage of ASIR Unlike FBP, ASIR identifies noise likely due to statistical fluctuations and uses that information to reconstruct a less noisy image. This is done by an iterative technique using image noise maps and axial images during reconstruction. The percentage of ASIR (10 100%) is operator selectable at the console. It reflects a linear 8 10 combination of the original FBP image (0% ASIR) and an essentially noise-free image created by full compliance with the mathematic model (100% ASIR). The same raw data collected for ASIR are fully available for FBP, and both sets of images are freely available for viewing. A choice of 40% ASIR implies that 40% of the ASIR image was blended with the FBP image. The percentage of ASIR is prescribed by the user, and multiple percentages of ASIRs can be reconstructed from the same scan data. Optimally, the percentage of ASIR would be customized for each scan because some patients may require a higher or lower percentage of ASIR on the basis of the amount of dose reduction and body habitus. Unfortunately, real-time variation in the percentage of ASIR (i.e., an auto-asir tab) for each examination is not time efficient because it is available only on the scanner console and not on an off-line workstation. At this time, determining the optimal percentage of ASIR for each patient would interrupt the scanning of other patients and disrupt workflow. For these reasons, in our practice we have preset the ASIR percentage for all scans, which are sent directly to the PACS. For the majority of low-dose abdominal CT scans, we use 40% ASIR, which was found to be well matched to the noise content of a full-dose examination with FBP in a previous study [14], as well as to physical measurements of the noise power spectrum (Fig. 1). The noise power spectrum [15] is generated by imaging a water phantom and plots the amount of noise (y-axis) against spatial frequency in line pairs (x-axis). Figure 1 compares the noise power spectrum of a water phantom imaged in three ways: full dose (25 mgy) with FBP, at half dose (12.5 mgy) with FBP, and at half dose (12.5 mgy) with 40% ASIR. The rationale for using 40% ASIR was based on the similarity of the half dose with 40% ASIR curve to the full dose with FBP curve. These two curves show similar noise at different frequencies with less noise than the half-dose FBP curve. Although the scanner Fig. 1 Noise power spectrum is a graphical depiction of image noise variance expressed as a function of its frequency. We derived the noise power spectrum using the method described by Boedeker and McNitt-Gray [15]. Noise power spectrum of American College of Radiology s routine reference dose of 25 mgy (solid thick line; with filtered back projection [FBP]) contains noise texture that is typical for abdominal images. Effect of reducing radiation dose by half to 12.5 mgy (dashed line; low-dose CT with FBP and 0% adaptive statistical iterative reconstruction [ASIR]) markedly increases noise across the full spectrum. Reconstruction of halfdose CT at 12.5 mgy with 40% ASIR (thin solid line) reduces noise, approximating the spectrum of full-dose FBP (thick line). 714 AJR:195, September 2010

Abdominal CT With Low Versus Routine Dose options included a choice of either 2D or 3D ASIR, we chose 2D ASIR because its reconstruction times were shorter than those for 3D ASIR. Routine-Dose CT With FBP Technique Comparison CT scans using routine-dose with FBP were obtained within an average of 244 days (range, 3 days to 4 years) of the low-dose CT scans with ASIR. The scan coverage (abdomen or abdomen and pelvis) and the enhancement phase (venous) matched the low-dose CT with ASIR for each patient. Routine-dose CT with FBP was performed on a 64-row (n = 32) or a 16-row (n = 21) scanner. Nine comparisons used the GE 64-row CT scanner (LightSpeed VCT, GE Healthcare), and 44 comparisons used Siemens Healthcare CT scanners (Volume Zoom, 16-row or 64-row). For peak kilovoltage, both 140 kvp (n = 45) and 120 kvp (n = 8) were used, and most of the examinations used dosemodulation software with variable milliamperage. The slice thickness was identical for both low-dose CT and routine-dose CT in 45% (24/53) of patients (5 mm for 16 patients and 3.75 mm for eight patients). The remaining 29 comparisons had slice thickness differences of less than 1 mm (low-dose with ASIR, 3.75 mm; routine-dose with FBP, 3 mm). Radiation Dose Measurements The volume CT dose indexes (CTDI vol ) for lowdose CT with ASIR and for routine-dose CT with FBP were obtained using our PACS (Centricity Radiology Solution, version 2.1, GE Healthcare). For every scanner used in this study, the accuracy of the manufacturer s displayed CTDI vol was tested as part of our routine quality control, and all of them agreed to within 10% of our independent measurements. The CTDI vol, dose length product, and effective dose in millisieverts (dose length product multiplied by a conversion factor of 0.015 [16]) for each examination was recorded for both the low-dose CT with ASIR and the routine-dose CT with FBP scans. The body mass index (BMI), calculated as weight in kilograms divided by height in meters squared, for each patient was recorded at the time of low-dose CT with ASIR. Both radiation dose and image quality measurements were compared in three patient groups determined by BMI: less than 20 (n = 7), 20 24.9 (n = 12), and 25 or greater (n = 34). Qualitative Analysis The 106 data sets (53 low-dose CT scans with ASIR and 53 routine-dose CT scans with FBP) were randomized and rendered anonymous so readers were unaware of which scanning technique had been used. Images were displayed on highresolution diagnostic monitors. All data sets were displayed at soft-tissue settings (window/level, approximately 400/40 HU). A radiologist who was not involved with grading the examinations selected representative images from each low-dose and routine-dose examination. The first set of images was matched to the same level of liver showing the main portal vein. These images were assessed for image noise, diagnostic acceptability, and artifacts (Fig. 2). The second set of images was matched to the same level of the aorta below the superior mesenteric artery to assess the image sharpness of the aortic wall. Independent qualitative image analysis was then performed by two board-certified and fellowshiptrained abdominal radiologists with 8 10 years of CT experience. Side-by-side comparison of lowdose CT images with ASIR and routine-dose CT images with FBP obtained from the same patients was performed with the scanning technique masked; comparison images were displayed on side-by-side monitors in a one-to-one format. The image display was randomly rotated so that the low-dose CT image was on the left monitor sometimes and on the right at other times. Image sharpness, image noise, and diagnostic acceptability were all graded on a scale from 1 (worst) to 5 (best) (Table 1). Image sharpness was assessed by evaluating aortic wall sharpness. Artifacts were quantified on a scale from 1 to 3: present and affecting image interpretation (grade 1), present but not affecting interpretation (grade 2), or absent (grade 3). A subset analysis of image quality results was performed for those patients who had identical slice thickness for both the low-dose and the routine-dose CT scans (n = 24). Quantitative Analysis In addition, one general imaging radiologist who was not involved in the qualitative data analysis obtained patient-specific quantitative A noise measurements for all 53 low-dose and all 53 routine-dose CT scans. Circular identically sized 200-mm 2 regions of interest were drawn in normal liver parenchyma and outside the body to measure image noise (as SDs of attenuation value). The liver parenchyma region of interest was positioned in the mid liver, excluding large hepatic vessels. Background noise measurements were made by recording the SD in a region of interest placed 5 mm outside the anterior abdominal wall at the same level. Statistical Analysis Grades for image sharpness, image noise, artifacts, and diagnostic acceptability were compared, calculated for each reader, and averaged across both readers. The average scores were compared using the paired Student s t test (SAS 9.0 software, SAS Institute); p < 0.01 was considered to indicate a statistically significant difference. Results Radiation Dose Low-dose CT with ASIR averaged 17 mgy CTDI vol compared with 25 mgy CTDI vol for routine-dose CT with FBP (Table 2). This represents an average lower dose of 33% (range, 2 75%) for the low-dose with ASIR protocol. For low-dose CT with ASIR, the average dose length product and effective dose were 860 mgy cm and 13 msv, respectively, compared with 1,193 mgy cm and 18 msv for routinedose CT with FBP. The average BMI was 26.8 (range, 17.8 43.9). When dose data were reviewed according to the BMI of patients, the percentage reduction in mean CTDI vol increased as BMI decreased. Therefore, patients with a low BMI (< 20) had lower doses (reduction Fig. 2 58-year-old man with body mass index (kg/m 2 ) of 26.5. Images in both panels matched same level of liver, which includes main portal vein and medial edge of posterior liver. A, Representative low-dose CT image with adaptive statistical iterative reconstruction (volume CT dose index [CTDI vol ], 9 mgy; 120 kv; 3.75 mm slice thickness) is shown. Note decreased sharpness of aortic wall. B, Representative routine-dose CT image with filtered back projection (CTDI vol, 18 mgy; 120 kvp; 3.75 mm slice thickness) is shown. B AJR:195, September 2010 715

Sagara et al. TABLE 1: Grading Scale for Qualitative Analysis of CT Examinations Qualitative Grading Scale in CTDI vol of 66% vs 23%) than did patients with a higher BMI ( 25). Image Quality Sharpness Noise Artifacts Diagnostic Acceptability 1 Blurry Unacceptable noise Present and affecting image interpretation Unacceptable 2 Poorer than average Above-average increased noise Present but not affecting interpretation Suboptimal 3 Average Average noise in an acceptable image Absent Average 4 Better than average Less-than-average noise Not applicable Above average 5 Sharpest Minimum or no noise Not applicable Superior TABLE 2: Volume CT Dose Index (CTDI vol ) and Image Quality Comparisons of Low-Dose CT With Adaptive Statistical Iterative Reconstruction (LDCT) Versus Routine-Dose CT With Filtered Back Projection (RDCT) by Body Mass Index (BMI) Patient Group, by BMI No. of Patients Average CTDI vol (mgy) Reduced Diagnostic LDCT RDCT CTDI vol (%) Sharpness b Noise b Artifacts c Acceptability b Qualitative Comparisons In spite of an average 33% lower radiation dose, low-dose CT with ASIR images were still graded as having significantly less image noise than routine-dose CT with FBP images (p < 0.001) (Table 2). Although routine-dose CT with FBP had superior grades for image sharpness (p < 0.001), diagnostic acceptability was not different than that for low-dose CT images with ASIR (p = 0.033). There was no significant difference in the presence of artifacts (p = 0.2). Image quality scores compared by BMI groups are also shown in Table 2. In larger patients (BMI, 25), low-dose CT images with ASIR were graded as significantly less noisy than routine-dose CT images with FBP (p < 0.001), similar to the overall analysis. In average-sized patients (BMI, 20 24.9), there was no significant difference in image noise. In smaller patients (BMI, < 20), however, there was a trend toward increased noise for lowdose CT with ASIR, although the trend was not statistically significant. Low-dose CT with ASIR had better diagnostic acceptability for larger patients and was not statistically different for average or small patients. For patients of all sizes, image sharpness for routine-dose CT with FBP was significantly better than that for low-dose CT with ASIR. No significant difference was found for artifacts on the basis of patient size. Subset Analysis of Examinations With Identical Slice Thickness Examination of 24 (45%) of 53 patients had an identical slice thickness for both low-dose and routine-dose CT. Most comparisons still had different peak kilovoltage (n = 15). Therefore, nine comparisons overall were performed with the same slice thickness and peak kilovoltage and using scanners from the same vendor. Seventeen patients had a BMI of 25 or greater, whereas the rest had a BMI of 20 24.9 (n = 5) or less than 20 (n = 2). The results for these 24 patients (Table 3) were similar to those for the overall analysis. For example, low-dose CT with ASIR had significantly decreased noise compared with routine-dose CT with FBP, but artifacts and diagnostic acceptability were not statistically different. The main difference in this subset was that there was no significant difference in image sharpness between the two groups. Image Quality of LDCT/Image Quality of RDCT a 25 34 21 27 23 2.7/2.8 3.2/2.6 d 2.9/2.9 3.1/2.9 d 20 24.9 12 13 23 44 2.3/2.8 d 2.8/2.8 2.9/2.8 2.8/2.9 < 20 7 7 21 66 1.9/2.8 d 2.5/2.8 3.0/2.9 2.9/3.0 All patients 53 17 25 33 2.5/2.8 e 3.0/2.7 e 2.9/2.9 3.0/2.9 a Superior results are in bold type. b Qualitative grading scale: 1 = unacceptable, 2 = suboptimal, 3 = average, 4 = above average, and 5 = superior. c Grading scale for artifacts: 1 = present and affecting image interpretation, 2 = present but not affecting interpretation, and 3 = absent. d p < 0.01. e p < 0.001. Quantitative Comparisons The average liver parenchymal noise measurements were significantly reduced using low-dose CT with ASIR (p < 0.001) (Table 4). Background noise was also lower using low-dose CT with ASIR compared with routine-dose CT with FBP, but this difference did not reach statistical significance. Discussion Image acquisition strategies to reduce dose include limiting scanning coverage, reducing multiphase protocols, improving scanner efficiency, increasing attenuating filtration, and altering scanning parameters, such as milliamperage and peak kilovoltage [7, 8, 11, 17 20]. Additional dose-reduction acquisition strategies include the use of automated tube modulation techniques [7, 18, 21] and the adoption of low-dose CT protocols for renal stone detection and CT colonography [22 25]. Noise from low-dose images can be reduced using postprocessing techniques, such as noise-reduction kernels [4, 6, 12, 26]; however, these kernels may also reduce diagnostic acceptability and lesion conspicuity [6]. Efforts to straightforwardly reduce the dose by simply lowering the milliamperage per second are severely hampered by image 716 AJR:195, September 2010

Abdominal CT With Low Versus Routine Dose TABLE 3: Subset Analysis of 24 CT Comparisons With Same Slice Thickness Measure of Image Quality Image Quality of LDCT/Image Quality of RDCT p Sharpness 2.6/2.8 0.08 Noise 3.1/2.8 0.01 Artifacts 2.8/2.9 0.02 Diagnostic Acceptability 3.0/2.9 0.5 Note LDCT = low-dose CT with adaptive statistical iterative reconstruction; RDCT = routine-dose CT with filtered back projection. degradation resulting from increased image noise. Image noise from this approach results from two sources: the expected noise resulting from limiting the number of photons and the reliance on FBP, the standard image reconstruction method. Although it is speedy in performance, FBP unfortunately always adds noise to the image as a result of its mathematics, because it mistakenly accepts data from each angled radiograph projection as being equally valid and fully free of noise. This erroneous assumption results in the extension of noise throughout the pixels in the slice, which is exceedingly difficult to reduce after incorporation in the image. ASIR is a newly available method that has the potential to provide improved image quality at lower radiation doses than FBP. ASIR starts with a normal FBP image and then creates progressive iterations to produce an essentially noise-free image (100% ASIR). These iterations work by modeling the noise in every projection. ASIR assumes that small noise differences between neighboring projections are valid and are due to the statistical nature of radiation. Larger differences are penalized by the model and lowered during the reconstruction process, thereby reducing image noise. The radiologist can specify the amount of ASIR incorporated into the final image by choosing a percentage from 10% to 100%. Mathematically, the ASIR percentage is simply a linear combination of the original FBP image and the full 100% ASIR image. In general, as the percentage of ASIR increases, the image noise decreases. In this study, the CTDI vol of using a low dose with ASIR protocol was, on average, 33% lower (and as much as 75% lower in some patients) compared with our routine dose with FBP protocol. This translated to an average effective dose reduction of 28% using the low-dose protocol. Despite this one-third reduction in the radiation dose, however, the overall image noise was lower both qualitatively and quantitatively when compared with routine-dose CT images from the same patients. The measured and observed image noise, however, varied by the size of the patient because of the more aggressive dose reduction that occurred in smaller patients. For example, in patients with a BMI less than 20, the average CTDI vol dose reduction was 66% versus 23% for patients with a BMI of 25 or greater. Not surprisingly, this dramatic dose reduction did result in both perceptible and measurable increases in image noise compared with routine-dose CT with FBP (Fig. 3), although this difference was not statistically significant and diagnostic acceptability scores remained nearly identical. Dose reductions using a fixed noise index in smaller patients may be too aggressive or at least proportionally larger compared with those for average-sized and large patients. This finding has also been reported in previous studies, which showed that the same noise level results in worse image quality for pediatric (small) versus obese patients [18, 27, 28], which may be due to the lack of fat planes in smaller patients. To minimize overly aggressive dose reductions that can occur from automatic dose modulation in smaller patients, we have subsequently increased the minimal accepted milliamperage from 100 to 200 ma (or 50 100 ma). This reduces the likelihood that small patients would be given a dose reduction of more than 50% compared with a routine-dose protocol. Image sharpness also appeared to be affected by changes in BMI, likely resulting from the aggressive dose reductions in smaller patients, as described earlier in the article. In this study, the worst image sharpness was present in smaller patients who underwent the greatest dose reductions. In the subset of patients with identical slice thickness, no difference in image sharpness was found, which may be because most patients in this group had a BMI of 25 or greater and, therefore, a relatively higher radiation dose. Decreased aortic wall sharpness was likely not associated with decreased diagnostic acceptability in this study because these examinations were not conducted specifically to evaluate the aorta. It is possible, however, that decreased wall sharpness observed at low doses (Figs. 2A and 3A) could be problematic if low-dose CT angiography was performed. In the solid organs, this degradation of image sharpness could compromise diagnostic accuracy if it caused a well-circumscribed lesion to appear less well defined or infiltrative. Such degradation was not observed in our study (note the similar appearance of the hepatic metastases in Figure 4); however, it was not specifically evaluated and should be studied in the future. The scores for diagnostic acceptability for the low-dose CT protocol were nearly identical to those for routine-dose CT with FBP. Qualitative assessment of artifacts was also nearly identical for both types of examinations. These results confirm the findings of our preliminary phantom analysis and clinical feasibility study in a different group of 12 patients [14]. Both that study and the current one showed that a low-dose CT with ASIR protocol can achieve significant reductions in radiation dose without TABLE 4: Quantitative Noise Measurements in Low-Dose CT With Adaptive Statistical Iterative Reconstruction (LDCT) Versus Routine-Dose CT With Filtered Back Projection (RDCT) Type of CT Noise Measurements, by BMI All Patients (n = 53) BMI 25 (n = 34) BMI = 20 24.9 (n = 12) BMI < 20 (n = 7) Liver Background Liver Background Liver Background Liver Background LDCT 12 29 11 25 12 39 13 29 RDCT 14 36 15 31 13 47 11 43 p < 0.001 0.02 < 0.001 0.03 0.3 0.4 0.02 0.1 Note BMI = body mass index (kg/m 2 ). AJR:195, September 2010 717

Sagara et al. compromising diagnostic acceptability compared with routine-dose CT with FBP. Although dose reductions achieved in larger patients using a low-dose CT with ASIR protocol are less than those in average or smaller patients, they are still beneficial especially since the number of obese patients in the United States continues to increase [29]. For patients with a BMI of 25 or greater, low-dose CT with ASIR images were still graded as having less image noise than routine-dose CT images with FBP, even though the CTDI vol was sometimes reduced by only 2%. This reduction in image noise implies that reconstructing with ASIR instead of FBP not only is helpful to decrease noise for low-dose studies but also may be beneficial for full-dose CT examinations in large patients whose CT scans are often inherently noisy. In addition, the averaged diagnostic acceptability scores were higher, although not statistically significant, for low-dose with ASIR in larger patients. A Fig. 3 70-year-old man with body mass index (kg/m 2 ) of 18.8. A, Low-dose CT image with adaptive statistical iterative reconstruction (volume CT dose index [CTDI vol ], 6 mgy; 120 kvp; 3.75 mm slice thickness) is shown. B, Routine-dose CT image with filtered back projection (CTDI vol, 16 mgy; 140 kvp; 3 mm slice thickness) is shown. Note increased image noise in liver and decreased sharpness of aortic wall in A compared with B; however, diagnostic image quality is still maintained. A Fig. 4 61-year-old man with body mass index (kg/m 2 ) of 27 and hepatic metastasis (arrows) due to colon cancer. A, Decreased image noise is seen in low-dose CT image with adaptive statistical iterative reconstruction (volume CT dose index [CTDI vol ],18 mgy; 120 kvp; 3.75 mm slice thickness). B, Routine-dose CT image with filtered back projection (CTDI vol, 27 mgy; 120 kvp; 3.75 mm slice thickness) is shown. Aside from decreased noise in liver in A, quality of images in both panels is otherwise nearly identical. A limitation of this study is that discrepancies in tube voltage (peak kilovoltage), slice thickness, enhancement timing, and changes in BMI can all also affect image quality. In this study, slice thickness differences within the same patient comparison were less than 1 mm. Nevertheless, we performed a subset analysis matching for slice thickness in 24 patients and found comparable image quality scores between the low- and routine-dose techniques. Therefore, the small slice thickness differences between the low- and routine-dose CT scans in this study did not likely affect the image quality scores. Another limitation of this retrospective study is that routine-dose CT scans were not all obtained with the same equipment as that used for the lowdose studies. Inherent differences in scanner attributes could affect results. The purpose of our study, however, was not to perform an isolated evaluation of different reconstruction algorithms from a single manufacturer. Instead, B B our goal was to evaluate our standard routinedose with FBP protocol against a new lowdose with ASIR protocol to determine whether this low-dose approach was clinically acceptable. Although an intramanufacturer comparison might also provide valuable insights, our study design instead reflected what is done in our practice and in many other practices that obtain CT scans on multiple different scanners from different manufacturers. On the basis of the results of this study, we have fully integrated ASIR into our clinical practice such that low-dose abdominal CT is now the standard default protocol for any scanner with the ASIR available. The results of this study have led us to adjust the minimal milliamperage per second to prevent overaggressive dose reductions in smaller patients, which can adversely affect image quality. All abdominal imagers in our practice interpret low-dose CT examinations using ASIR. Although it is possible to change the percentage of ASIR, it is rarely, if ever, altered from our standard of 40 50% when used routinely. The availability of ASIR has increased the availability and use of lowdose CT protocols in our practice. There remains, however, much more work to be done to evaluate the application of these low-dose protocols. For example, diagnostic accuracy, the use in other body parts, and investigation of different iterative techniques that allow even greater dose reductions still remain to be assessed. The results of this study show that low-dose abdominal CT with ASIR is a viable technique with image quality that is nearly comparable to that of our routine-dose techniques and is worthy of further study. References 1. IMV Medical Information Division, Inc. IMV 2006 CT market summary report. www.imvinfo. com/user/documents/content_documents/nws_ rad/ms_ct_dsandtoc.pdf. Published 2006. Accessed September 16, 2009. 2. IMV Medical Information Division, Inc. IMV 2007 CT market summary report. www.imvinfo. com/user/documents/content_documents/def_ dis/2008_06_12_06_45_57_706.pdf. Published 2007. Accessed September 16, 2009. 3. Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med 2007; 357:2277 2284 4. Funama Y, Awai K, Miyazaki O, et al. Improvement of low-contrast detectability in low-dose hepatic multidetector computed tomography using a novel adaptive filter: evaluation with a computer-simulated liver including tumors. Invest Radiol 2006; 41:1 7 718 AJR:195, September 2010

Abdominal CT With Low Versus Routine Dose 5. Funama Y, Awai K, Nakayama Y, et al. Radiation dose reduction without degradation of low-contrast detectability at abdominal multisection CT with a low-tube voltage technique: phantom study. Radiology 2005; 237:905 910 6. Kalra MK, Maher MM, Blake MA, et al. Detection and characterization of lesions on low-radiation-dose abdominal CT images postprocessed with noise reduction filters. Radiology 2004; 232:791 797 7. Kalra MK, Maher MM, Toth TL, et al. Strategies for CT radiation dose optimization. Radiology 2004; 230:619 628 8. Linton OW, Mettler FA Jr. National conference on dose reduction in CT, with an emphasis on pediatric patients. AJR 2003; 181:321 329 9. Nakayama Y, Awai K, Funama Y, et al. Abdominal CT with low tube voltage: preliminary observations about radiation dose, contrast enhancement, image quality, and noise. Radiology 2005; 237:945 951 10. Okumura M, Ota T, Kainuma K, Sayre JW, Mc- Nitt-Gray M, Katada K. Effect of edge-preserving adaptive image filter on low-contrast detectability in CT systems: application of ROC analysis. Int J Biomed Imaging 2008; 2008:379486 11. Valentin J. International Commission on Radiation Protection. Managing patient dose in multidetector computed tomography (MDCT): ICRP publication 102. Ann ICRP 2007; 37:1 79 12. Kalra MK, Maher MM, Sahani DV, et al. Lowdose CT of the abdomen: evaluation of image improvement with use of noise reduction filters pilot study. Radiology 2003; 228:251 256 13. Thibault JB, Sauer KD, Bouman CA, Hsieh J. A three-dimensional statistical approach to improved image quality for multislice helical CT. Med Phys 2007; 34:4526 4544 14. Hara AK, Paden RG, Silva AC, Kujak JL, Lawder HJ, Pavlicek W. Iterative reconstruction technique for reducing body radiation dose at CT: feasibility study. AJR 2009; 193:764 771 [Erratum in AJR 2009; 193:1190] 15. Boedeker KL, McNitt-Gray MF. Application of the noise power spectrum in modern diagnostic MDCT. Part II. Noise power spectra and signal to noise. Phys Med Biol 2007; 52:4047 4061 16. American Association of Physicists in Medicine. The measurement, reporting, and management of radiation dose in CT: report of AAPM Task Group 23 of the Diagnostic Imaging Council CT Committee. AAPM report no. 96. College Park (MD): American Association of Physicists in Medicine. Published 2008. Accessed December 15, 2009. www.aapm.org/pubs/reports/rpt_96.pdf. 17. Lee CH, Goo JM, Ye HJ, et al. Radiation dose modulation techniques in the multidetector CT era: from basics to practice. RadioGraphics 2008; 28:1451 1459 18. McCollough CH, Bruesewitz MR, Kofler JM Jr. CT dose reduction and dose management tools: overview of available options. RadioGraphics 2006; 26:503 512 19. Mulkens TH, Bellinck P, Baeyaert M, et al. Use of an automatic exposure control mechanism for dose optimization in multi-detector row CT examinations: clinical evaluation. Radiology 2005; 237:213 223 20. Smith AB, Dillon WP, Lau BC, et al. Radiation dose reduction strategy for CT protocols: successful implementation in neuroradiology section. Radiology 2008; 247:499 506 21. Kalra MK, Maher MM, Toth TL, Kamath RS, Halpern EF, Saini S. Radiation from extra images acquired with abdominal and/or pelvic CT: effect of automatic tube current modulation. Radiology 2004; 232:409 414 22. Cohnen M, Vogt C, Beck A, et al. Feasibility of MDCT colonography in ultra-low-dose technique in the detection of colorectal lesions: comparison with high-resolution video colonoscopy. AJR 2004; 183:1355 1359 23. Luz O, Buchgeister M, Klabunde M, et al. Evaluation of dose exposure in 64-slice CT colonography. Eur Radiol 2007; 17:2616 2621 24. Niemann T, Kollmann T, Bongartz G. Diagnostic performance of low-dose CT for the detection of urolithiasis: a meta-analysis. AJR 2008; 191:396 401 25. Poletti PA, Platon A, Rutschmann OT, Schmidlin FR, Iselin CE, Becker CD. Low-dose versus standard-dose CT protocol in patients with clinically suspected renal colic. AJR 2007; 188:927 933 26. Funama Y, Awai K, Miyazaki O, et al. Radiation dose reduction in hepatic multidetector computed tomography with a novel adaptive noise reduction filter. Radiat Med 2008; 26:171 177 27. Wilting JE, Zwartkruis A, van Leeuwen MS, Timmer J, Kamphuis AG, Feldberg M. A rational approach to dose reduction in CT: individualized scan protocols. Eur Radiol 2001; 11:2627 2632 28. Kalra MK, Maher MM, Blake MA, Toth TL, Saini SS. Multidetector CT scanning of abdomen and pelvis: a study for optimization of automatic tube current modulation technique in 120 subjects (abstr). RSNA 2003. Oak Brook, IL: Radiological Society of North America, 2003 29. Ford ES, Mokdad AH. Epidemiology of obesity in the Western Hemisphere. J Clin Endocrinol Metab 2008; 93[suppl 1]:S1 S8 AJR:195, September 2010 719