- Email: [email protected]
Dose-modified 256-MDCT of the Abdomen Using Low Tube Current and Hybrid Iterative Reconstruction
Dose-modified 256-MDCT of the Abdomen Using Low Tube Current and Hybrid Iterative Reconstruction Jorge M. Fuentes-Orrego, MD, Koichi Hayano, MD, PhD, Avinash R. Kambadakone, MD, Peter F. Hahn, MD, PhD, Dushyant V. Sahani, MD Rationale and Objectives: To evaluate the performance of hybrid iterative reconstruction technique (h-IRT) on image quality (IQ) in abdominal dose-modified (DM) scans in phantom and in patients in comparison to filtered back projection (FBP). Materials and Methods: An anthropomorphic phantom was scanned using various kVp (80–140) and mAs (25–100) settings. Images were reconstructed with FBP and h-IRT levels (1–6). In 69 adults (59.6 13.54 years; 20 male, 49 female), DM computed tomography (CT) scans were performed using 120 kVp and 100–120 mAs. In 25/69, 5-mm FBP and h-IRT (levels 1–4 and 5) images were analyzed to validate IQ. The subsequent 44/69 had FBP and h-IRT (level 4) images reconstructed. Two readers evaluated 188 image series for IQ, noise, and artifacts. Objective and subjective data were analyzed using t-test and Wilcoxon signed-rank test, respectively. In 46/69 patients, prior dose CT was available for dose comparison. Results: In the phantom, noise reduction ranged from 12% (h-IRT level 1) to 50% (level 6). In patients, h-IRT level 4 images were rated diagnostic in 69/69 exams but DM-FBP images were found nondiagnostic in 20/69 patients. The size-specific dose estimate (SSDE) was reduced by 55% in the dose-modified CT group, (SSDE:4.55 1.15 mGy) over the prior dose protocol (SSDE:10.21 3.5 mGy, P < .0001). Conclusion: h-IRT improved IQ in abdominal DM-CT scans in phantom and in patients. Dose improvements were greater in smaller patients than larger ones. Key Words: Radiation dose; abdomen; computed tomography; iterative reconstruction; image quality. ªAUR, 2013
S
ince the introduction of CT scanners, technical advances in hardware and software have made its implementation possible in a variety of clinical conditions (1). These benefits of computed tomography (CT) imaging have also introduced new challenges of increased radiation exposure to the patients (2–4). The CT manufacturers and users have focused mainly on a tube current time product (mAs)-based approach to lower dose because of a predictable linearity between radiation dose and tube current and its straightforward application in practice. However, the lower image quality on the traditional filtered back projection (FBP) reconstructed scans performed using a substantially reduced dose remains a major limitation of dose-modified scans (5). Major CT manufacturers have introduced iterative reconstruction techniques (IRTs) to reduce noise and artifact to improve image quality (6–11). Acad Radiol 2013; 20:1405–1412 From the Division of Abdominal Imaging and Intervention, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, 55 Fruit Street, Boston, MA 02115 (J.M.F.-O., K.H., A.R.K., P.F.H., D.V.S.). Received July 10, 2013; accepted August 6, 2013. Address correspondence to: D.V.S. e-mail: [email protected] ªAUR, 2013 http://dx.doi.org/10.1016/j.acra.2013.08.004
Hybrid IRT (h-IRT, iDose4) functions in the projection domain and in image space for suppressing image noise and artifacts and improving geometry through a complex mathematical algorithm. The noisiest signals in the areas of low photon counts are identified first in the projection domain and suppressed by the iterative reconstruction process; remaining noise is then propagated in the image domain, where it is easy to locate and remove (12). To our knowledge, this reconstruction technique has been studied on CT scans of the abdomen acquired with a low tube voltage (80-kV) approach in smaller sized adult patients and shown the promising results of decreasing both radiation dose and contrast media while preserving image quality (12,13). Tube voltage (kV) has a quadratic relationship with the dose; its modification therefore achieves greater dose savings than mAs having more linearity with the dose. Low kV is often applied in high-contrast examinations such as CT angiographies, CT urography, arterial phase acquisitions, or kidney stone evaluation, where higher image noise can be accepted. However, low kV is not routinely practiced in low-contrast studies such as portal venous phase scans because of image quality concerns and inadequate familiarity with optimal kV selection and corresponding tube current modifications to preserve the diagnostic quality in patients of medium to larger 1405
FUENTES-ORREGO ET AL
Academic Radiology, Vol 20, No 11, November 2013
Figure 1. Axial images of lower torso phantom acquired at 120 kVp and 50 mAs, reconstructed using filtered back projection (a) and hybrid iterative reconstruction level 4 (b), at the level of L3 showing a loop of colon with two polyps. Observe reduction in the graininess (image noise) of image (b) in comparison to image (a) after applying hybrid iterative reconstruction. A region of interest of similar size and location was manually drawn in the extracolonic fat tissue in both images. Note a similar attenuation value on both images; in contrast, the objective noise (standard deviation of computed tomography number) has decreased from 12.72 (a) to 9.082 (b) after applying hybrid iterative reconstruction level 4.
body size (14). Therefore, the purpose of our study was to investigate the performance of h-IRT on image quality in dose-modified CT protocols using a low mAs approach in a phantom and in patients in comparison to FBP.
MATERIALS AND METHODS All images were acquired on a 256-slice CT scanner (Brilliance iCT; Philips Healthcare, Cleveland Ohio). Rationale for Choosing the Level of IRT
Phantom study. To assess the impact each level of iterative reconstruction had on objective noise along various dose settings, a phantom study was performed first. We used a custom anthropomorphic 35-cm lower torso phantom (Phantom Laboratory, Salem, NY) with materials representing extracolonic fat, colon, and pelvic bones. The phantom was built using a polymer to simulate abdominal tissues radiographically, both in morphology and attenuation. This phantom has been described in more detail in a prior report and it was considered to be a good surrogate for a medium sized adult patient (15) (Fig 1). The phantom was placed supine on the CT table to simulate our standard abdominal CT scanning technique. We applied various CT settings: four kVp settings (140, 120, 100, and 80 kVp) and four mAs settings (25, 50, 75, and 100 reference mAs) for each kVp selection. One of the phantom acquisitions matched our objective modified protocol (120 kVp, 100 mAs) to assess objective noise among various levels of iterative reconstruction. The automatic tube current modulation system (Z-DOM; Philips Healthcare) was applied as practiced in our clinical protocol. This technique modulates the tube current along the longitudinal scanning direction of the body size and attenuation profile of the region being scanned to obtain a constant image quality. Each one of the acquisitions was reconstructed at the scanner console using the FBP technique and the six levels of h-IRT (1–6). Objective noise measurements were performed manually by 1406
drawing an oval region of interest (ROI) averaging 423 mm2 (range 420–430 mm2) in the simulated extracolonic fat tissue density region to estimate the standard deviation (SD). Three measurements at different locations in the extracolonic fat tissue density area were taken at the same image level to avoid imprecision; the mean value of SD was computed for statistical analysis. An average noise reduction in the range of 12% (level 1) to 50% (level 6) was noted with each increment in h-IRT level from 1 to 6 (P < .001). The noise suppression was greater at lower dose parameters (80 kVp and 25 mAs) compared to higher doses (140 kVp and 75 mAs) and reached a plateau at higher dose settings. For the purpose of this study, only noise measurements at 100 and 120 kVp are shown (Table 1). Patient study. A total of 69 consecutive adult patients who underwent CT scans of the abdomen and pelvis with positive oral (900 mL, Readi CAT, containing 1–2% w/v of barium sulfate; Bracco Diagnostics, Princeton, NJ) and intravenous contrast in the portal venous phase during May 2012 were included in the study (mean age, 59.61 13.53 years; age range, 29–88 years; mean weight, 73.85 14.84 kg; 20 men [mean age, 60.52 13.67 years] and 49 women [mean age, 59.22 13.6 years]). Intravenous access was placed in the upper extremity using a 20-gauge catheter. Based on each patient’s body weight (<60 kg, 61–90 kg, and >91 kg) 80, 100, or 120 mL of a nonionic contrast medium (Isovue, 370 mg of iodine/mL; Bracco Diagnostics) was bolus injected intravenously using a mechanical power injector (Medrad, Bayer-Schering, Berlin, Germany) at a rate of 3 mL/sec. The most common clinical indications for the abdominal scans were varied and included: change in bowel habits, unexplained weight loss, suspicion or follow-up of abdominal tumors, and chronic abdominal pain, among others. The most common imaging findings were: positive for tumor (n = 37, colon, lymphoma, hepatocellular carcinoma, renal, ovarian cancer, other), metastatic disease (n = 7, lung, breast, prostate, melanoma, peritoneal implants, other), inflammatory process (n = 5, diverticulitis, Crohn’s disease, acute cholecystitis), findings unrelated to patient’s symptoms
Academic Radiology, Vol 20, No 11, November 2013
DOSE-MODIFIED CT SCANS OF THE ABDOMEN USING LOW TUBE CURRENT AND HYBRID ITERATIVE RECONSTRUCTION
TABLE 1. Noise Measurements Obtained in the Lower Torso Phantom at 100 and 120 kVp Using Different Tube Current-time Product Settings for FBP and h-IRT Images 100 kVp Technique FBP h-IRT 1 h-IRT 2 h-IRT 3 h-IRT 4 h-IRT 5 h-IRT 6
120 kVp 25 mAs
50 mAs
75 mAs
100 mAs
25 mAs
50 mAs
75 mAs
100 mAs
30.4 8.88 21.1 3.40 20.5 3.69 19.0 3.36 17.6 3.28 15.8 3.21 14.0 2.54
19.5 4.8 16.0 3.1 15.3 3.0 14.1 3.0 12.2 2.8 11.6 2.4 10.7 2.6
15.9 3.4 13.8 2.4 12.8 2.2 11.8 2.1 10.9 1.9 10.1 1.8 8.87 1.6
14.3 2.9 12.2 2.6 11.7 2.1 10.9 2.0 10.0 1.9 9.09 1.9 8.06 1.4
19.7 4.2 15.3 1.9 13.9 1.6 12.9 1.5 12.1 1.8 11.0 1.4 9.93 1.1
14.9 3.9 12.8 3.1 12.0 2.7 11.1 2.8 10.3 2.5 9.38 2.1 8.64 2.2
12.7 2.8 11.1 2.1 10.1 1.8 9.33 1.6 8.73 1.6 8.18 1.5 7.10 1.5
11.3 2.1 10.1 2.0 9.67 1.9 9.21 1.8 8.70 1.8 7.66 1.6 6.95 1.4
FBP, filtered back projection; h-IRT, hybrid iterative reconstruction technique. Scores are means standard deviations.
(n = 11, fatty liver, cholelithiasis, fibroid uterus, cysts [liver, renal, and adnexal], hepatic hemangioma, other), and no abnormality (n = 9). The exclusion criteria for our study were any of the following: patients <18 years old, multiple phase contrast-enhanced CT scan of the abdomen, history of intravenous contrast allergies, renal insufficiency or at risk for contrast-induced nephropathy, and subjects with poor intravenous access. CT technique. We have been continuously optimizing protocols through mAs modification taking a graded approach with feedback from radiologists in our division for image quality acceptability and image noise. We have had a CT system from the same vendor but without iterative reconstruction software application for more than 2 years and the routine abdominal scans with a reference milliamperage second (ref mAs) of 150–200 (150 mAs when body weight [BW] < 90 kg; 200 mAs for BW > 90 kg). Based on our previous experience with the FBP images from the same vendor and using iterative reconstruction techniques from other vendors for more than 3 years as well as the quantitative data from our phantom study, a reference mAs of 100–120 was selected based on patients’ body weight, reducing the dose in a range of 40–50% (100 mAs when BW < 90 kg; 120 mAs for BW > 90 kg). We anticipated image quality degradation for image series rendered using FBP with an estimated noise increment of 40% and we consciously decided to use hybrid iterative reconstruction level 4 as recommended by the CT manufacturer to improve image quality in the dosemodified scans. In addition, to evaluate the performance of this reconstruction algorithm, we decided to generate additional image series with other levels of iterative reconstruction as described below. The dose-modified CT protocol (DM-CT) was acquired as follows. All patients were positioned with their feet first on the CT table. The scanning extended craniocaudally from above the dome of the diaphragm to the symphysis pubis. The scanning was performed using 120 kVp, 0.5 seconds rotation time, reference mAs of 100–120 (automatic current selection) with an effective mAs ranging from 75 to 350 per section, 0.65 slice collimation, and 0.9 pitch. A fixed time delay of 70 seconds
following the initiation of contrast medium injection was used to trigger scanning. In 46 of 69 patients, prior standard dose CECT (PD-CT) scan was available for dose comparison. The PD-CT scanning parameters were 120 kVp, 0.5 seconds rotation time, effective mAs ranging from 104 to 517 per section, pitch of 0.9, with axial images reconstructed with 5-mm slice thickness. Image reconstruction. In the first 25 subjects, images of 5-mm thickness were reconstructed axially using FBP and three different hybrid reconstruction levels (1, 4, and 5) using a standard soft-tissue filter. In the subsequent 44 patients, images were generated using FBP and hybrid reconstruction level 4. In all of the scans, images were also reconstructed in the coronal and sagittal plane using a 3-mm thickness on the scanner console by the technologist. The axial and coronal image series were transferred to a digital picture archiving and communication system (PACS) diagnostic workstations (Impax RS 3000 review station; AGFA Technical Imaging Systems, Richfield Park, NJ). The axial images were used to assess noise, artifacts, and overall image quality on PACS. Image Analysis
Quantitative evaluation. A total of 188 CT image series was evaluated. Objective evaluation of image quality was performed by a single investigator (J.M.F.), with 5 years of experience in interpreting abdominal CTs. Objective noise measurements were made as follows. Noise was recorded at three different locations by manually placing an oval ROI with an average area of 1.53 cm2 (range, 1.50–1.60 cm2) within the fat in the anterior abdominal wall, mesenteric or retroperitoneal fat (eg, depending on the amount of fat, presence of ascites, mesenteric fat stranding), and within the posterior wall of the abdomen. The ROI placement was done at the level of lower pole of right kidney and was carefully placed to be similar in location in all the image series. The mean SD of the three ROIs was calculated. Additionally, to evaluate the impact body size has on image quality, the maximum transverse diameter (MTD) was measured on the scout image on every patient from skin to skin and 1407
FUENTES-ORREGO ET AL
the maximal anteroposterior diameter (APD) was measured on the axial CT images. To calculate the size-specific dose estimate, the MTD was used as described in the following section. Total BW was reviewed from the chart and entered into the database. Qualitative analysis. Qualitative analysis was performed 3 weeks apart by two board-certified and fellowship-trained abdominal radiologists (D.V.S. and A.R.K.) with 17 and 7 years of abdominal CT experience, respectively, at a PACS workstation in two different sessions. The 188 DM-CT image series was deidentified and codified to blind the readers from the reconstruction technique and scan parameters. The axial images were presented randomly at soft-tissue image viewing settings (window width 400 Hounsfield units [HU], window level 40 HU). However, the readers had full liberty to adjust image contrast brightness and magnification as desired. We followed the European Guidelines on Quality Criteria for Computed Tomography (16) that has also been applied by other investigators (17) for subjective assessment of image parameters such as noise, artifacts, and overall image quality (IQ). Because the abdominal CT scans assessed had been performed for various clinical indications, and in several patients pathologies were identified in multiple organs without a reference standard validation for every lesion type, diagnostic performance of CT images was not evaluated. Instead, reader’s acceptance of overall image quality served as the standard for assessing the diagnostic confidence. Image noise was rated on a four-grade scale (grade 1 = absence of noise; grade 2 = minor noise; grade 3 = average noise; and grade 4 = above average, unacceptable for diagnostic interpretation). Artifacts such as streak and beam hardening were rated on scale of 4 (grade 1 = absence of artifacts; grade 2 = minor artifacts, diagnosis is certain; grade 3 = major artifacts, limit structure visualization, diagnosis is still possible; and grade 4 = major artifacts, diagnosis not possible). The overall image quality was scored on 5-grade scale (grade 1 = poor, unacceptable; grade 2 = below average, diagnosis not possible; grade 3 = acceptable, diagnosis is possible; grade 4 = good, diagnosis is certain; and grade 5 = excellent). Any case was considered nondiagnostic when either of the radiologists rated it as unacceptable (if IQ score was <3). Radiation dose measurements. Volume CT dose index (CTDI) and dose-length product (DLP) were recorded using the dose information page provided by the scanner. Additionally, the effective diameter (ED) and the size-specific dose estimate (SSDE) were calculated using the patient’s MTD following the American Association of Physicists in Medicine report 204 (18). Statistical analysis. Data analysis was performed using Statview (SAS Institute, Inc., Cary, NC). The subjective image noise, artifacts, and overall IQ ratings for DM-CT scans reconstructed with FBP and h-IRT were compared using Wilcoxon signed-rank test. Noise measurements (SD of CT number) were obtained for FBP and hybrid reconstructed 1408
Academic Radiology, Vol 20, No 11, November 2013
images. Additionally, anthropometric data including BW, MTD, APD, and CTDI, DLP, and mean tube current time products were also averaged both in the dose-modified and prior dose CT scans. All these data were compared using multiple paired t-tests. A Spearman’s correlation test was performed between BW and SSDE. Interobserver variability was calculated using Cohen’s k statistics and percentage agreement between the two radiologists for each of the subjective parameters evaluated. Quality of agreement was defined as follows: k < 0.20, poor agreement; k = 0.2–0.4, fair agreement; k = 0.41–0.60, moderate agreement; k = 0.61–0.8, good agreement; and k = 0.81–1.00, excellent agreement, following interpretations in prior studies (19). The patient categorization based on BW into average size to smaller (BW < 90 kg) or heavier (BW $ 91 kg) was derived from our institutional practice of basing CT protocols on this dichotomy. This study was Health Insurance Portability and Accountability Act–compliant. The Institutional Review Board approved and waived informed consent for this retrospective study. RESULTS There was a significant difference in patient BW among the 21 male (mean weight 87.82 kg 21.41; mean MTD 44.95 3.16 cm; mean APD 28.73 2.63 cm) and 49 female subjects (74.75 kg 13.92; mean MTD 44.42 3.99 cm; mean APD 25.17 3.17 cm) (P < .0001). No such difference was found for MTD and APD. Quantitative and Qualitative Image Analysis
Hybrid reconstructed images using levels 4 and 5 were considered diagnostic in all 25 patients and received higher image quality scores than FBP (P < .0001). In contrast, images rendered using level 1 showed no difference in the IQ score and noise when compared to FBP. Given that images generated with hybrid reconstruction levels 4 and 5 had comparable IQ scores (P > .05), we applied level 4 on the remaining 44 patients along with FBP, with the latter serving as a reference for comparison. Noise measurements, as expected, were highest for FBP images and decreased as the influence of h-IRT increased. In the 69 patients, unacceptable image noise (score $ 3) was rated on two FBP images by one of the readers in the qualitative analysis, whereas both readers scored the remaining FBP and hybrid level four images for minor noise (score < 3) in the remaining patients (Fig 2)(Table 2). The hybrid reconstructed images were rated diagnostic in all 69 subjects, but in 20 patients (29%) FBP images were deemed not of diagnostic quality (P < .001). Subjective IQ ratings were supported by the quantitative analysis because objective noise measurements on diagnostic FBP images were significantly lower over nondiagnostic FBP (17.01 2.79 vs. 19.26 3.17; P = .0045). The anthropometric variables had no impact on image quality of scans rendered using hybrid reconstruction. However, a modest difference was found for
Academic Radiology, Vol 20, No 11, November 2013
DOSE-MODIFIED CT SCANS OF THE ABDOMEN USING LOW TUBE CURRENT AND HYBRID ITERATIVE RECONSTRUCTION
Figure 2. Axial contrast enhanced multidetector row computed tomography obtained at the level of the liver (window width, 301 Hounsfield units; window level 44) in a 65-year-old female patient with multiple hypoattenuating metastatic lesions in the liver. Dose-modified images (a–d) acquired at 120 kVp, 100 reference mAs, and size-specific dose estimate 6.86 mGy. (a) Filtered back projection, (b) hybrid iterative reconstruction (h-IRT) level 1, (c) h-IRT level 4, and (d) h-IRT level 5. Note that streak artifacts and noise decrease at levels 4 and 5 of hybrid reconstruction demonstrating preserved conspicuity of a subcentimeter hypoattenuating lesion (white arrow) in segment VI. At level 5 of iterative reconstruction (d), there is an increased pixilated appearance of iterative reconstructed images.
TABLE 2. Averaged Scores for Subjective Image Quality for Both Readers (R1/R2)
Reconstruction Technique FBP h-IRT level 1 h-IRT level 4 h-IRT level 5 FBP h-IRT level 4
Minor Noise (Score <3)
Minor Artifact (Score <3)
Acceptable Overall Image Quality (Score >3)
R1, 1.78 R2, 1.75 R1, 1.46 R2, 1.55 R1, 1.15 R2, 1.09 R1, 1.03 R2, 1.09 R1, 1.87 R2, 1.75 R1, 1.13 R2, 1.17
R1, 1.25 R2, 1.23 R1, 1.19 R2, 1.23 R1, 1.02 R2, 1.23 R1, 1.02 R2, 1.19 R1, 1.28 R2, 1.13 R1, 1.03 R2, 1.08
R1, 3.00 R2, 3.2 R1, 3.12 R2, 3.32 R1, 3.50 R2, 3.9 R1, 3.92 R2, 3.96 R1, 2.91 R2, 3.15 R1, 3.62 R2, 3.74
FBP, filtered back projection; h-IRT, hybrid iterative reconstruction technique. Images were considered diagnostic quality if subjective score for noise/artifacts <3, and overall IQ was >3.
Figure 3. Axial dose-modified computed tomography obtained at 120 kVp and 120 mAs, size-specific dose estimate 9.12 mGy (window width, 451 Hounsfield units [HU]; window level, 102 HU) in a 36-year-old woman with mesenteric fibrosis (white arrow). Image quality on filtered back projection (a) was rated suboptimal because of increased noise. After applying hybrid reconstruction level 4 (b), overall image quality was considered acceptable for diagnostic interpretation.
the mean APD between diagnostic and nondiagnostic FBP images because patients whose APD > 26 cm had reduced IQ scores (P = .04)(Fig 3).
The readers’ interobserver agreement ranged from fair to modest (k = 0.2–0.5) for subjective noise and overall image quality, and from minor to fair (k = 0.01–0.4) for image 1409
FUENTES-ORREGO ET AL
Academic Radiology, Vol 20, No 11, November 2013
Figure 4. Axial and coronal contrast enhanced multidetector row computed tomography obtained at the level of the pelvis (window width, 329 Hounsfield units; window level 53) in a 23-year-old female patient with history of Crohn’s disease showing a loop of terminal ileum with wall thickening and mucosal enhancement (white arrow). Dose-modified images (a–c) acquired at 120 kVp, 100 reference mAs, and size-specific dose estimate 4.84 mGy reconstructed with hybrid iterative reconstruction level 4. Images (b–d) were acquired with our prior dose protocol at 120 kVp, 200 reference mAs, and sizespecific dose estimate 10.4 mGy. The iterative reconstruction has preserved image quality at 53% reduced dose settings in this patient.
artifacts. When the readers’ ratings were grouped as acceptable (rating $ 3) and unacceptable (<3) for overall image quality and subjective noise and artifacts scores, the agreement for different image parameters ranged from 71% to 100%. For subjective noise, both readers agreed 97% (67/69) when FBP images were evaluated, and it reached 100% (69/69) in the remaining reconstructed image series. Similarly, ratings were concordant between both readers in 99% (68/69) when artifacts were assessed in all FBP images and achieved 100% for the remaining hybrid reconstructed image series. Slightly more variability was found when overall image quality was scored when readers agreed 71% (49/69) on FBP images; 92% (23/25) for h-IRT level 1, 100% (69/69) for h-IRT level 4, and 100% (25/25) h-IRT level 5, respectively (Fig 4). Radiation Doses
Overall, the DM-CT scans reduced the DLP by 50% and the CTDI by 55% relative to the PD-CT scans in the 46/69 patients that had a prior study. The mean radiation dose in PD-CT scan (CTDIvol: 11.43 4.13 mGy, DLP: 526.3 234.37 mGy-cm, ED: 38.74 2.9 cm, SSDE: 10.21 3.5 mGy) was 55% higher than DM-CT scan group (CTDIvol: 5.11 1.35 mGy, DLP: 263.21 110.8 mGy-cm, ED: 38.6 2.52 cm, SSDE: 4.55 1.15 mGy). A positive correlation was found between the patients BW and SSDE (P = .004, r2 = 0.115). Finding that most of 1410
the subjects (55/69) in our study were average to small sized (BW < 90 kg) based on our BW categorization, we chose the median (BW = 72 kg) as the cutoff value to analyze the data. The dose reduction was higher in smaller size patients (BW # 72 kg, 62.17 7.63 kg; SSDE, 4.23 17.04 mGy vs. BW > 72 kg, 86.59 9.45 kg; SSDE, 4.92 21.5mSv; P = .014). DISCUSSION Radiation exposure from CT continues to remain a patient safety concern, especially in patients of small body size and younger age (20). Measures have been taken by both radiologists and CT manufacturers to lower radiation dose for various body parts, and the abdomen is no exception (21,22). Despite the reported success in dose reduction, efforts to lower dose further have raised concerns over compromised examination quality in images reconstructed using the traditional FBP technique. To overcome these inherent flaws in the FBP process, major CT manufacturers have introduced iterative reconstruction algorithms to improve image quality (6,12,23–25). iDose4 is a recently introduced hybrid reconstruction technique that works both in the projection and image domains by identifying and correcting signals that are likely to produce noise-related artifacts by applying a photon statistics algorithm (13). This entire process occurs at sufficient speed up to 20 images per second (26); images appear during each examination in almost real time. We have
ASIR, adaptive statistical iterative reconstruction; CTDI, volume computed tomography dose index; DLP, dose-length product; IRIS, image reconstruction in space.
ASIR ASIR ASIR ASIR IRIS iDose 26.8/— 24/— 21.3/127 —/135.08 19.1/121.2 —/162.5 860 470 306 380 182.8 263 17.0 12.0 6.15 7.7 3.5 5.11 1,193 894 1054 577 365.7 526 25.0 22.0 20.79 12 7.0 11.43 Noise index selection noise index selection Low kVp (80) Noise index selection Low mAs (100) Low mAs (100–120) Sagara (29) Hara (27) Kaza (28) Kambadakone (30) Lee (5) Authors study
Mean Dose Modified DLP (mGy-cm) Mean Dose Modified CTDI (mGy) Mean Previous DLP (mGy-cm) Mean Previous Dose CTDI (mGy) Dose Reduction Technique Author
TABLE 3. Dose Reduction on Dose-modified Abdominal CECT Scans Using Different Reconstruction Techniques
Mean Body Mass Index/Body Weight (lb)
Iterative Reconstruction Technique
Academic Radiology, Vol 20, No 11, November 2013
DOSE-MODIFIED CT SCANS OF THE ABDOMEN USING LOW TUBE CURRENT AND HYBRID ITERATIVE RECONSTRUCTION
demonstrated a 55% dose reduction that does not compromise diagnostic quality when hybrid IRT is used to reconstruct the CT images. Conventional FBP failed to generate diagnostic quality images in 29% of the same DM-CT scan image series. The clinical claims of h-IRT for improving image quality in abdominal CT exams performed at 33% to 70% lower dose than FBP protocols have been reported with other reconstruction algorithms (5,27–30). However, to appreciate dose improvements better and compare radiation dose more accurately between these studies, more quantitative measures such as absolute CTDI or DLP in mGy-cm are appropriate. Our mean CTDI values for patients weighing less and more than 90 kg were 5.0 mGy and 5.5 mGy, respectively, and these doses are at the lower end of the spectrum of abdomen CT dose from other published studies in patients with similar body size (5,27–30) (Table 3). In a recent publication, Nakaura et al reported very encouraging results with the use of hybrid iterative reconstruction in multiphasic liver CT at low kVp settings (13). The authors demonstrated the feasibility of reducing both dose and iodinated contrast medium use by 50% and 40%, respectively, while preserving image quality and contrast-to-noise ratio with hybrid iterative reconstruction in small size adults. Although using low kVp is desirable for its dual benefit of improving image contrast and lowering CT dose, its application in portal venous phase examinations is still evolving (14). Patient body size in North America remains a challenge with low kVp use. This is further compounded by incomplete understanding of appropriate kVp and mA selection for each patient to meet image quality expectations (31). But low kVp use has been more popular for CT examinations with high intrinsic contrast such as arterial phase acquisitions and CT angiography (32–36). Moreover, given the recent introduction of this hybrid reconstruction technique in our practice, we have implemented a graded approach to gain experience, familiarize the radiologists with the image quality, and learn about any potential limitations in interpreting dose-modified CT exams. Indeed, we have discovered that the influence level of h-IRT affects the quality of CT images because DM-CT data reconstructed with h-IRT level 1 were rated as nondiagnostic quality in 12% of the 25 patients to which it was applied. Our phantom study results are similar to observations made by other investigators using the Philips hybrid reconstruction algorithm and other partial iterative reconstruction approaches. iDose allows significant improvement in image noise over FBP regardless of the influence level of hybrid reconstruction applied (5,26,35). Our study has a few limitations. First, the sample size is relatively small to reliably assess the impact of anthropometric variables and different body sizes on image quality. We did not encounter enough patients with very large body habitus. Moreover, our study did not evaluate the utility of this hybrid reconstruction on lesion conspicuity and diagnosis of various abnormalities in the abdomen because we have studied patients with varying clinical indications. However, the main objective of assessing the performance this iterative 1411
FUENTES-ORREGO ET AL
reconstruction technique had on image quality in dosemodified CT scans of the abdomen was achieved. Indeed, we have used image quality and reader confidence to assess radiologist acceptance of these images in rendering diagnostic interpretations. Finally, the possible effect of using low tube potential settings on lowering radiation dose and the impact of h-IRT on image quality was not investigated in this pilot study because we do not practice low-kVp acquisitions for portal venous phase exams. In conclusion, our study shows that hybrid iterative reconstruction technique (iDose4) can be reliably applied to improve image quality and noise in dose-modified abdominal CECT scans. Using our described approach, a 55% drop in the CT radiation dose can be accomplished over our previous dose FBP images, which could be advantageous to all patients, especially benefitting smaller and averaged sized patients.
REFERENCES 1. Boone JM. Multidetector CT opportunities, challenges, and concerns associated with scanners with 64 or more detector rows. Radiology 2006; 241:334–337. 2. Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med 2007; 357:2277–2284. lez A, Mahesh M, Kim K, et al. Projected cancer risks 3. Berrington de Gonza from computed tomographic scans performed in the United States in 2007. Arch Intern Med 2009; 169:2071–2077. 4. Thrall JH. Radiation exposure in CT scanning and risk: where are we? Radiology 2012; 264:325–328. 5. Lee SJ, Park SH, Kim AY, et al. A prospective comparison of standarddose CT enterography and 50% reduced-dose CT enterography with and without noise reduction for evaluating Crohn disease. AJR Am J Roentgenol 2011; 197:50–57. 6. Fleischmann D, Boas FE. Computed tomography—old ideas and new technology. Eur Radiol 2011; 21:510–517. €ller HC, et al. Iterative reconstruction 7. Schindera ST, Diedrichsen L, Mu algorithm for abdominal multidetector CT at different tube voltages: assessment of diagnostic accuracy, image quality, and radiation dose in a phantom study. Radiology 2011; 260:454–462. 8. Marin D, Nelson RC, Schindera ST, et al. Low-tube-voltage, high-tubecurrent multidetector abdominal CT: improved image quality and decreased radiation dose with adaptive statistical iterative reconstruction algorithm—initial clinical experience. Radiology 2010; 254:145–153. 9. Prakash P, Kalra MK, Kambadakone AK, et al. Reducing abdominal CT radiation dose with adaptive statistical iterative reconstruction technique. Invest Radiol 2010; 45:202–210. 10. Pickhardt PJ, Lubner MH, Kim DH, et al. Abdominal CT with model-based iterative reconstruction (MBIR): initial results of a prospective trial comparing ultralow-dose with standard-dose imaging. AJR Am J Roentgenol 2012; 199:1–9. €ger M, Dahlstro € m N, et al. Radiation dose reduction 11. Kalra MK, Woisetschla with sinogram affirmed iterative reconstruction technique for abdominal computed tomography. J Comput Assist Tomogr 2012; 36:339–346. 12. Nakaura T, Awaki K, Maruyama N, et al. Abdominal dynamic CT in patients with renal dysfunction: contrast agent dose reduction with low tube voltage and high tube current-time product settings at 256-detector row CT. Radiology 2011; 261:467–476. 13. Nakaura T, Nakamura S, Maruyama N, et al. Low contrast agent and radiation dose protocol for hepatic dynamic CT of thin adults at 256-detector row CT: effect of low tube voltage and hybrid iterative reconstruction algorithm on image quality. Radiology 2012; 264:445–454. ~es LS, Fletcher JG, Harmsen WS, et al. Appropriate patient selec14. Guimara tion at abdominal dual-energy CT using 80 kV: relationship between patient size, image noise, and image quality. Radiology 2010; 257:732–742. 15. Zalis ME, Perumpillichira JJ, Kim JY, et al. Polyp Size at CT colonography after electronic subtraction cleansing in an anthropomorphic colon phantom. Radiology 2005; 236:118–124.
1412
Academic Radiology, Vol 20, No 11, November 2013
16. European Commission European guidelines on quality criteria for computed tomography. (EUR 16262 EN). Luxembourg: European Commission & The Office For Official Publications of the European Communities, 2000. www. drs.dk/guidelines/ct/quality/Download.htm. Accessed July 10, 2013. 17. Singh S, Kalra MK, Hsieh J, et al. Abdominal CT: comparison of adaptive statistical iterative and filtered back projection reconstruction techniques. Radiology 2010; 257:373–383. 18. American Association of Physicists in Medicine. Size-specific dose estimates (SSDE) in pediatric and adult body CT examinations. Task Group 204. College Park, MD: American Association of Physicists in Medicine, 2011. 19. Tack D, Bohy P, Perlot I, et al. Suspected acute colon diverticulitis: imaging with low-dose unenhanced multi-detector row CT. Radiology 2005; 237: 189–196. 20. Lee SH, Kim MJ, Yoon CS, et al. Radiation dose reduction with the adaptive statistical iterative reconstruction (ASIR) technique for chest CT in children: an intra-individual comparison. Eur J Radiol 2012; 81:e938–e943, http:// www.sciencedirect.com/science/article/pii/S0720048X1200280X. 21. Liu W, Esler SJ, Kenny BJ, et al. Low-dose non enhanced helical CT of renal colic: assessment of ureteric stone detection and measurement of effective dose equivalent. Radiology 2000; 215:51–54. 22. Marin D, Nelson RC, Barnhart H, et al. Detection of pancreatic tumors, image quality, and radiation dose during the pancreatic parenchymal phase: effect of a low-tube-voltage, high-tube-current CT technique—preliminary results. Radiology 2010; 256:450–459. 23. Kulkarni NM, Uppot RN, Eisner BH, et al. Radiation dose reduction at multidetector CT with adaptive statistical iterative reconstruction for evaluation of urolithiasis: how low can we go? Radiology 2012; 265:158–166. 24. Baker ME, Dong F, Primak A, et al. Contrast-to-noise ratio and lowcontrast object resolution on full- and low-dose MDCT: SAFIRE versus filtered back projection in a low-contrast object phantom and in the liver. AJR Am J Roentgenol 2012; 199:8–18. 25. Negi N, Yoshikawa T, Ohno Y, et al. Hepatic CT perfusion measurements: A feasibility study for radiation dose reduction using new image reconstruction method. Eur J Radiol 2012; 81:3048–3054. €l PB, Fingerle AA, Renger B, et al. Initial performance characterization 26. Noe of a clinical noise-suppressing reconstruction algorithm for MDCT. AJR Am J Roentgenol 2011; 197:1404–1409. 27. Hara AK, Paden RG, Silva AC, et al. Iterative reconstruction technique for reducing body radiation dose at CT: feasibility study. AJR Am J Roentgenol 2009; 193:764–771. 28. Kaza RK, Platt JF, Al-Hawary MH, et al. CT enterography at 80 kVp with adaptive statistical iterative reconstruction versus at 120 kVp with standard reconstruction: image quality, diagnostic adequacy and dose reduction. AJR Am J Roentgenol 2012; 198:1084–1092. 29. Sagara Y, Hara A, Pavlicek W, et al. Abdominal CT: comparison of low dose CT with adaptive statistical iterative reconstruction and routine dose CT with filtered back projection in 53 patients. AJR Am J Roentgenol 2010; 195:713–719. 30. Kambadakone AR, Chaudhary NA, Desai GS, et al. Low dose MDCT and CT enterography of patients with Crohn disease: feasibility of adaptive statistical iterative reconstruction. AJR Am J Roentgenol 2011; 196: W743–W752. 31. Yu L, Li H, Fletcher JG, et al. Automatic selection of tube potential for radiation dose reduction in CT: a general strategy. Med Phys 2010; 37: 234–243. 32. Iezzi R, Cotroneo AR, Giammarino A, et al. Low-dose multidetector-row CT-angiography of abdominal aortic aneurysm after endovascular repair. Eur J Radiol 2011; 79:21–28. 33. Sahani DV, Kalva SP, Hahn PF, et al. 16-MDCT angiography in living kidney donors at various tube potentials: impact on image quality and radiation dose. AJR Am J Roentgenol 2007; 188:115–120. 34. Namimoto T, Oda S, Utsunomiya D, et al. Improvement of image quality at low-radiation dose and low-contrast material dose abdominal CT in patients with cirrhosis: intraindividual comparison of low tube voltage with iterative reconstruction algorithm and standard tube voltage. J Comput Assist Tomogr 2012; 36:495–501. 35. Kalva SP, Sahani DV, Hahn PF, et al. Using the K-edge to improve contrast conspicuity and to lower radiation dose with a 16-MDCT: a phantom and human study. J Comput Assist Tomogr 2006; 30:391–397. 36. Pinho DF, Kulkarni NM, Krishnaraj A, et al. Initial experience with singlesource dual-energy CT abdominal angiography and comparison with single-energy CT angiography: image quality, enhancement, diagnosis and radiation dose. Eur J Radiol 2013; 23:351–359.
S
ince the introduction of CT scanners, technical advances in hardware and software have made its implementation possible in a variety of clinical conditions (1). These benefits of computed tomography (CT) imaging have also introduced new challenges of increased radiation exposure to the patients (2–4). The CT manufacturers and users have focused mainly on a tube current time product (mAs)-based approach to lower dose because of a predictable linearity between radiation dose and tube current and its straightforward application in practice. However, the lower image quality on the traditional filtered back projection (FBP) reconstructed scans performed using a substantially reduced dose remains a major limitation of dose-modified scans (5). Major CT manufacturers have introduced iterative reconstruction techniques (IRTs) to reduce noise and artifact to improve image quality (6–11). Acad Radiol 2013; 20:1405–1412 From the Division of Abdominal Imaging and Intervention, Department of Radiology, Massachusetts General Hospital, Harvard Medical School, 55 Fruit Street, Boston, MA 02115 (J.M.F.-O., K.H., A.R.K., P.F.H., D.V.S.). Received July 10, 2013; accepted August 6, 2013. Address correspondence to: D.V.S. e-mail: [email protected] ªAUR, 2013 http://dx.doi.org/10.1016/j.acra.2013.08.004
Hybrid IRT (h-IRT, iDose4) functions in the projection domain and in image space for suppressing image noise and artifacts and improving geometry through a complex mathematical algorithm. The noisiest signals in the areas of low photon counts are identified first in the projection domain and suppressed by the iterative reconstruction process; remaining noise is then propagated in the image domain, where it is easy to locate and remove (12). To our knowledge, this reconstruction technique has been studied on CT scans of the abdomen acquired with a low tube voltage (80-kV) approach in smaller sized adult patients and shown the promising results of decreasing both radiation dose and contrast media while preserving image quality (12,13). Tube voltage (kV) has a quadratic relationship with the dose; its modification therefore achieves greater dose savings than mAs having more linearity with the dose. Low kV is often applied in high-contrast examinations such as CT angiographies, CT urography, arterial phase acquisitions, or kidney stone evaluation, where higher image noise can be accepted. However, low kV is not routinely practiced in low-contrast studies such as portal venous phase scans because of image quality concerns and inadequate familiarity with optimal kV selection and corresponding tube current modifications to preserve the diagnostic quality in patients of medium to larger 1405
FUENTES-ORREGO ET AL
Academic Radiology, Vol 20, No 11, November 2013
Figure 1. Axial images of lower torso phantom acquired at 120 kVp and 50 mAs, reconstructed using filtered back projection (a) and hybrid iterative reconstruction level 4 (b), at the level of L3 showing a loop of colon with two polyps. Observe reduction in the graininess (image noise) of image (b) in comparison to image (a) after applying hybrid iterative reconstruction. A region of interest of similar size and location was manually drawn in the extracolonic fat tissue in both images. Note a similar attenuation value on both images; in contrast, the objective noise (standard deviation of computed tomography number) has decreased from 12.72 (a) to 9.082 (b) after applying hybrid iterative reconstruction level 4.
body size (14). Therefore, the purpose of our study was to investigate the performance of h-IRT on image quality in dose-modified CT protocols using a low mAs approach in a phantom and in patients in comparison to FBP.
MATERIALS AND METHODS All images were acquired on a 256-slice CT scanner (Brilliance iCT; Philips Healthcare, Cleveland Ohio). Rationale for Choosing the Level of IRT
Phantom study. To assess the impact each level of iterative reconstruction had on objective noise along various dose settings, a phantom study was performed first. We used a custom anthropomorphic 35-cm lower torso phantom (Phantom Laboratory, Salem, NY) with materials representing extracolonic fat, colon, and pelvic bones. The phantom was built using a polymer to simulate abdominal tissues radiographically, both in morphology and attenuation. This phantom has been described in more detail in a prior report and it was considered to be a good surrogate for a medium sized adult patient (15) (Fig 1). The phantom was placed supine on the CT table to simulate our standard abdominal CT scanning technique. We applied various CT settings: four kVp settings (140, 120, 100, and 80 kVp) and four mAs settings (25, 50, 75, and 100 reference mAs) for each kVp selection. One of the phantom acquisitions matched our objective modified protocol (120 kVp, 100 mAs) to assess objective noise among various levels of iterative reconstruction. The automatic tube current modulation system (Z-DOM; Philips Healthcare) was applied as practiced in our clinical protocol. This technique modulates the tube current along the longitudinal scanning direction of the body size and attenuation profile of the region being scanned to obtain a constant image quality. Each one of the acquisitions was reconstructed at the scanner console using the FBP technique and the six levels of h-IRT (1–6). Objective noise measurements were performed manually by 1406
drawing an oval region of interest (ROI) averaging 423 mm2 (range 420–430 mm2) in the simulated extracolonic fat tissue density region to estimate the standard deviation (SD). Three measurements at different locations in the extracolonic fat tissue density area were taken at the same image level to avoid imprecision; the mean value of SD was computed for statistical analysis. An average noise reduction in the range of 12% (level 1) to 50% (level 6) was noted with each increment in h-IRT level from 1 to 6 (P < .001). The noise suppression was greater at lower dose parameters (80 kVp and 25 mAs) compared to higher doses (140 kVp and 75 mAs) and reached a plateau at higher dose settings. For the purpose of this study, only noise measurements at 100 and 120 kVp are shown (Table 1). Patient study. A total of 69 consecutive adult patients who underwent CT scans of the abdomen and pelvis with positive oral (900 mL, Readi CAT, containing 1–2% w/v of barium sulfate; Bracco Diagnostics, Princeton, NJ) and intravenous contrast in the portal venous phase during May 2012 were included in the study (mean age, 59.61 13.53 years; age range, 29–88 years; mean weight, 73.85 14.84 kg; 20 men [mean age, 60.52 13.67 years] and 49 women [mean age, 59.22 13.6 years]). Intravenous access was placed in the upper extremity using a 20-gauge catheter. Based on each patient’s body weight (<60 kg, 61–90 kg, and >91 kg) 80, 100, or 120 mL of a nonionic contrast medium (Isovue, 370 mg of iodine/mL; Bracco Diagnostics) was bolus injected intravenously using a mechanical power injector (Medrad, Bayer-Schering, Berlin, Germany) at a rate of 3 mL/sec. The most common clinical indications for the abdominal scans were varied and included: change in bowel habits, unexplained weight loss, suspicion or follow-up of abdominal tumors, and chronic abdominal pain, among others. The most common imaging findings were: positive for tumor (n = 37, colon, lymphoma, hepatocellular carcinoma, renal, ovarian cancer, other), metastatic disease (n = 7, lung, breast, prostate, melanoma, peritoneal implants, other), inflammatory process (n = 5, diverticulitis, Crohn’s disease, acute cholecystitis), findings unrelated to patient’s symptoms
Academic Radiology, Vol 20, No 11, November 2013
DOSE-MODIFIED CT SCANS OF THE ABDOMEN USING LOW TUBE CURRENT AND HYBRID ITERATIVE RECONSTRUCTION
TABLE 1. Noise Measurements Obtained in the Lower Torso Phantom at 100 and 120 kVp Using Different Tube Current-time Product Settings for FBP and h-IRT Images 100 kVp Technique FBP h-IRT 1 h-IRT 2 h-IRT 3 h-IRT 4 h-IRT 5 h-IRT 6
120 kVp 25 mAs
50 mAs
75 mAs
100 mAs
25 mAs
50 mAs
75 mAs
100 mAs
30.4 8.88 21.1 3.40 20.5 3.69 19.0 3.36 17.6 3.28 15.8 3.21 14.0 2.54
19.5 4.8 16.0 3.1 15.3 3.0 14.1 3.0 12.2 2.8 11.6 2.4 10.7 2.6
15.9 3.4 13.8 2.4 12.8 2.2 11.8 2.1 10.9 1.9 10.1 1.8 8.87 1.6
14.3 2.9 12.2 2.6 11.7 2.1 10.9 2.0 10.0 1.9 9.09 1.9 8.06 1.4
19.7 4.2 15.3 1.9 13.9 1.6 12.9 1.5 12.1 1.8 11.0 1.4 9.93 1.1
14.9 3.9 12.8 3.1 12.0 2.7 11.1 2.8 10.3 2.5 9.38 2.1 8.64 2.2
12.7 2.8 11.1 2.1 10.1 1.8 9.33 1.6 8.73 1.6 8.18 1.5 7.10 1.5
11.3 2.1 10.1 2.0 9.67 1.9 9.21 1.8 8.70 1.8 7.66 1.6 6.95 1.4
FBP, filtered back projection; h-IRT, hybrid iterative reconstruction technique. Scores are means standard deviations.
(n = 11, fatty liver, cholelithiasis, fibroid uterus, cysts [liver, renal, and adnexal], hepatic hemangioma, other), and no abnormality (n = 9). The exclusion criteria for our study were any of the following: patients <18 years old, multiple phase contrast-enhanced CT scan of the abdomen, history of intravenous contrast allergies, renal insufficiency or at risk for contrast-induced nephropathy, and subjects with poor intravenous access. CT technique. We have been continuously optimizing protocols through mAs modification taking a graded approach with feedback from radiologists in our division for image quality acceptability and image noise. We have had a CT system from the same vendor but without iterative reconstruction software application for more than 2 years and the routine abdominal scans with a reference milliamperage second (ref mAs) of 150–200 (150 mAs when body weight [BW] < 90 kg; 200 mAs for BW > 90 kg). Based on our previous experience with the FBP images from the same vendor and using iterative reconstruction techniques from other vendors for more than 3 years as well as the quantitative data from our phantom study, a reference mAs of 100–120 was selected based on patients’ body weight, reducing the dose in a range of 40–50% (100 mAs when BW < 90 kg; 120 mAs for BW > 90 kg). We anticipated image quality degradation for image series rendered using FBP with an estimated noise increment of 40% and we consciously decided to use hybrid iterative reconstruction level 4 as recommended by the CT manufacturer to improve image quality in the dosemodified scans. In addition, to evaluate the performance of this reconstruction algorithm, we decided to generate additional image series with other levels of iterative reconstruction as described below. The dose-modified CT protocol (DM-CT) was acquired as follows. All patients were positioned with their feet first on the CT table. The scanning extended craniocaudally from above the dome of the diaphragm to the symphysis pubis. The scanning was performed using 120 kVp, 0.5 seconds rotation time, reference mAs of 100–120 (automatic current selection) with an effective mAs ranging from 75 to 350 per section, 0.65 slice collimation, and 0.9 pitch. A fixed time delay of 70 seconds
following the initiation of contrast medium injection was used to trigger scanning. In 46 of 69 patients, prior standard dose CECT (PD-CT) scan was available for dose comparison. The PD-CT scanning parameters were 120 kVp, 0.5 seconds rotation time, effective mAs ranging from 104 to 517 per section, pitch of 0.9, with axial images reconstructed with 5-mm slice thickness. Image reconstruction. In the first 25 subjects, images of 5-mm thickness were reconstructed axially using FBP and three different hybrid reconstruction levels (1, 4, and 5) using a standard soft-tissue filter. In the subsequent 44 patients, images were generated using FBP and hybrid reconstruction level 4. In all of the scans, images were also reconstructed in the coronal and sagittal plane using a 3-mm thickness on the scanner console by the technologist. The axial and coronal image series were transferred to a digital picture archiving and communication system (PACS) diagnostic workstations (Impax RS 3000 review station; AGFA Technical Imaging Systems, Richfield Park, NJ). The axial images were used to assess noise, artifacts, and overall image quality on PACS. Image Analysis
Quantitative evaluation. A total of 188 CT image series was evaluated. Objective evaluation of image quality was performed by a single investigator (J.M.F.), with 5 years of experience in interpreting abdominal CTs. Objective noise measurements were made as follows. Noise was recorded at three different locations by manually placing an oval ROI with an average area of 1.53 cm2 (range, 1.50–1.60 cm2) within the fat in the anterior abdominal wall, mesenteric or retroperitoneal fat (eg, depending on the amount of fat, presence of ascites, mesenteric fat stranding), and within the posterior wall of the abdomen. The ROI placement was done at the level of lower pole of right kidney and was carefully placed to be similar in location in all the image series. The mean SD of the three ROIs was calculated. Additionally, to evaluate the impact body size has on image quality, the maximum transverse diameter (MTD) was measured on the scout image on every patient from skin to skin and 1407
FUENTES-ORREGO ET AL
the maximal anteroposterior diameter (APD) was measured on the axial CT images. To calculate the size-specific dose estimate, the MTD was used as described in the following section. Total BW was reviewed from the chart and entered into the database. Qualitative analysis. Qualitative analysis was performed 3 weeks apart by two board-certified and fellowship-trained abdominal radiologists (D.V.S. and A.R.K.) with 17 and 7 years of abdominal CT experience, respectively, at a PACS workstation in two different sessions. The 188 DM-CT image series was deidentified and codified to blind the readers from the reconstruction technique and scan parameters. The axial images were presented randomly at soft-tissue image viewing settings (window width 400 Hounsfield units [HU], window level 40 HU). However, the readers had full liberty to adjust image contrast brightness and magnification as desired. We followed the European Guidelines on Quality Criteria for Computed Tomography (16) that has also been applied by other investigators (17) for subjective assessment of image parameters such as noise, artifacts, and overall image quality (IQ). Because the abdominal CT scans assessed had been performed for various clinical indications, and in several patients pathologies were identified in multiple organs without a reference standard validation for every lesion type, diagnostic performance of CT images was not evaluated. Instead, reader’s acceptance of overall image quality served as the standard for assessing the diagnostic confidence. Image noise was rated on a four-grade scale (grade 1 = absence of noise; grade 2 = minor noise; grade 3 = average noise; and grade 4 = above average, unacceptable for diagnostic interpretation). Artifacts such as streak and beam hardening were rated on scale of 4 (grade 1 = absence of artifacts; grade 2 = minor artifacts, diagnosis is certain; grade 3 = major artifacts, limit structure visualization, diagnosis is still possible; and grade 4 = major artifacts, diagnosis not possible). The overall image quality was scored on 5-grade scale (grade 1 = poor, unacceptable; grade 2 = below average, diagnosis not possible; grade 3 = acceptable, diagnosis is possible; grade 4 = good, diagnosis is certain; and grade 5 = excellent). Any case was considered nondiagnostic when either of the radiologists rated it as unacceptable (if IQ score was <3). Radiation dose measurements. Volume CT dose index (CTDI) and dose-length product (DLP) were recorded using the dose information page provided by the scanner. Additionally, the effective diameter (ED) and the size-specific dose estimate (SSDE) were calculated using the patient’s MTD following the American Association of Physicists in Medicine report 204 (18). Statistical analysis. Data analysis was performed using Statview (SAS Institute, Inc., Cary, NC). The subjective image noise, artifacts, and overall IQ ratings for DM-CT scans reconstructed with FBP and h-IRT were compared using Wilcoxon signed-rank test. Noise measurements (SD of CT number) were obtained for FBP and hybrid reconstructed 1408
Academic Radiology, Vol 20, No 11, November 2013
images. Additionally, anthropometric data including BW, MTD, APD, and CTDI, DLP, and mean tube current time products were also averaged both in the dose-modified and prior dose CT scans. All these data were compared using multiple paired t-tests. A Spearman’s correlation test was performed between BW and SSDE. Interobserver variability was calculated using Cohen’s k statistics and percentage agreement between the two radiologists for each of the subjective parameters evaluated. Quality of agreement was defined as follows: k < 0.20, poor agreement; k = 0.2–0.4, fair agreement; k = 0.41–0.60, moderate agreement; k = 0.61–0.8, good agreement; and k = 0.81–1.00, excellent agreement, following interpretations in prior studies (19). The patient categorization based on BW into average size to smaller (BW < 90 kg) or heavier (BW $ 91 kg) was derived from our institutional practice of basing CT protocols on this dichotomy. This study was Health Insurance Portability and Accountability Act–compliant. The Institutional Review Board approved and waived informed consent for this retrospective study. RESULTS There was a significant difference in patient BW among the 21 male (mean weight 87.82 kg 21.41; mean MTD 44.95 3.16 cm; mean APD 28.73 2.63 cm) and 49 female subjects (74.75 kg 13.92; mean MTD 44.42 3.99 cm; mean APD 25.17 3.17 cm) (P < .0001). No such difference was found for MTD and APD. Quantitative and Qualitative Image Analysis
Hybrid reconstructed images using levels 4 and 5 were considered diagnostic in all 25 patients and received higher image quality scores than FBP (P < .0001). In contrast, images rendered using level 1 showed no difference in the IQ score and noise when compared to FBP. Given that images generated with hybrid reconstruction levels 4 and 5 had comparable IQ scores (P > .05), we applied level 4 on the remaining 44 patients along with FBP, with the latter serving as a reference for comparison. Noise measurements, as expected, were highest for FBP images and decreased as the influence of h-IRT increased. In the 69 patients, unacceptable image noise (score $ 3) was rated on two FBP images by one of the readers in the qualitative analysis, whereas both readers scored the remaining FBP and hybrid level four images for minor noise (score < 3) in the remaining patients (Fig 2)(Table 2). The hybrid reconstructed images were rated diagnostic in all 69 subjects, but in 20 patients (29%) FBP images were deemed not of diagnostic quality (P < .001). Subjective IQ ratings were supported by the quantitative analysis because objective noise measurements on diagnostic FBP images were significantly lower over nondiagnostic FBP (17.01 2.79 vs. 19.26 3.17; P = .0045). The anthropometric variables had no impact on image quality of scans rendered using hybrid reconstruction. However, a modest difference was found for
Academic Radiology, Vol 20, No 11, November 2013
DOSE-MODIFIED CT SCANS OF THE ABDOMEN USING LOW TUBE CURRENT AND HYBRID ITERATIVE RECONSTRUCTION
Figure 2. Axial contrast enhanced multidetector row computed tomography obtained at the level of the liver (window width, 301 Hounsfield units; window level 44) in a 65-year-old female patient with multiple hypoattenuating metastatic lesions in the liver. Dose-modified images (a–d) acquired at 120 kVp, 100 reference mAs, and size-specific dose estimate 6.86 mGy. (a) Filtered back projection, (b) hybrid iterative reconstruction (h-IRT) level 1, (c) h-IRT level 4, and (d) h-IRT level 5. Note that streak artifacts and noise decrease at levels 4 and 5 of hybrid reconstruction demonstrating preserved conspicuity of a subcentimeter hypoattenuating lesion (white arrow) in segment VI. At level 5 of iterative reconstruction (d), there is an increased pixilated appearance of iterative reconstructed images.
TABLE 2. Averaged Scores for Subjective Image Quality for Both Readers (R1/R2)
Reconstruction Technique FBP h-IRT level 1 h-IRT level 4 h-IRT level 5 FBP h-IRT level 4
Minor Noise (Score <3)
Minor Artifact (Score <3)
Acceptable Overall Image Quality (Score >3)
R1, 1.78 R2, 1.75 R1, 1.46 R2, 1.55 R1, 1.15 R2, 1.09 R1, 1.03 R2, 1.09 R1, 1.87 R2, 1.75 R1, 1.13 R2, 1.17
R1, 1.25 R2, 1.23 R1, 1.19 R2, 1.23 R1, 1.02 R2, 1.23 R1, 1.02 R2, 1.19 R1, 1.28 R2, 1.13 R1, 1.03 R2, 1.08
R1, 3.00 R2, 3.2 R1, 3.12 R2, 3.32 R1, 3.50 R2, 3.9 R1, 3.92 R2, 3.96 R1, 2.91 R2, 3.15 R1, 3.62 R2, 3.74
FBP, filtered back projection; h-IRT, hybrid iterative reconstruction technique. Images were considered diagnostic quality if subjective score for noise/artifacts <3, and overall IQ was >3.
Figure 3. Axial dose-modified computed tomography obtained at 120 kVp and 120 mAs, size-specific dose estimate 9.12 mGy (window width, 451 Hounsfield units [HU]; window level, 102 HU) in a 36-year-old woman with mesenteric fibrosis (white arrow). Image quality on filtered back projection (a) was rated suboptimal because of increased noise. After applying hybrid reconstruction level 4 (b), overall image quality was considered acceptable for diagnostic interpretation.
the mean APD between diagnostic and nondiagnostic FBP images because patients whose APD > 26 cm had reduced IQ scores (P = .04)(Fig 3).
The readers’ interobserver agreement ranged from fair to modest (k = 0.2–0.5) for subjective noise and overall image quality, and from minor to fair (k = 0.01–0.4) for image 1409
FUENTES-ORREGO ET AL
Academic Radiology, Vol 20, No 11, November 2013
Figure 4. Axial and coronal contrast enhanced multidetector row computed tomography obtained at the level of the pelvis (window width, 329 Hounsfield units; window level 53) in a 23-year-old female patient with history of Crohn’s disease showing a loop of terminal ileum with wall thickening and mucosal enhancement (white arrow). Dose-modified images (a–c) acquired at 120 kVp, 100 reference mAs, and size-specific dose estimate 4.84 mGy reconstructed with hybrid iterative reconstruction level 4. Images (b–d) were acquired with our prior dose protocol at 120 kVp, 200 reference mAs, and sizespecific dose estimate 10.4 mGy. The iterative reconstruction has preserved image quality at 53% reduced dose settings in this patient.
artifacts. When the readers’ ratings were grouped as acceptable (rating $ 3) and unacceptable (<3) for overall image quality and subjective noise and artifacts scores, the agreement for different image parameters ranged from 71% to 100%. For subjective noise, both readers agreed 97% (67/69) when FBP images were evaluated, and it reached 100% (69/69) in the remaining reconstructed image series. Similarly, ratings were concordant between both readers in 99% (68/69) when artifacts were assessed in all FBP images and achieved 100% for the remaining hybrid reconstructed image series. Slightly more variability was found when overall image quality was scored when readers agreed 71% (49/69) on FBP images; 92% (23/25) for h-IRT level 1, 100% (69/69) for h-IRT level 4, and 100% (25/25) h-IRT level 5, respectively (Fig 4). Radiation Doses
Overall, the DM-CT scans reduced the DLP by 50% and the CTDI by 55% relative to the PD-CT scans in the 46/69 patients that had a prior study. The mean radiation dose in PD-CT scan (CTDIvol: 11.43 4.13 mGy, DLP: 526.3 234.37 mGy-cm, ED: 38.74 2.9 cm, SSDE: 10.21 3.5 mGy) was 55% higher than DM-CT scan group (CTDIvol: 5.11 1.35 mGy, DLP: 263.21 110.8 mGy-cm, ED: 38.6 2.52 cm, SSDE: 4.55 1.15 mGy). A positive correlation was found between the patients BW and SSDE (P = .004, r2 = 0.115). Finding that most of 1410
the subjects (55/69) in our study were average to small sized (BW < 90 kg) based on our BW categorization, we chose the median (BW = 72 kg) as the cutoff value to analyze the data. The dose reduction was higher in smaller size patients (BW # 72 kg, 62.17 7.63 kg; SSDE, 4.23 17.04 mGy vs. BW > 72 kg, 86.59 9.45 kg; SSDE, 4.92 21.5mSv; P = .014). DISCUSSION Radiation exposure from CT continues to remain a patient safety concern, especially in patients of small body size and younger age (20). Measures have been taken by both radiologists and CT manufacturers to lower radiation dose for various body parts, and the abdomen is no exception (21,22). Despite the reported success in dose reduction, efforts to lower dose further have raised concerns over compromised examination quality in images reconstructed using the traditional FBP technique. To overcome these inherent flaws in the FBP process, major CT manufacturers have introduced iterative reconstruction algorithms to improve image quality (6,12,23–25). iDose4 is a recently introduced hybrid reconstruction technique that works both in the projection and image domains by identifying and correcting signals that are likely to produce noise-related artifacts by applying a photon statistics algorithm (13). This entire process occurs at sufficient speed up to 20 images per second (26); images appear during each examination in almost real time. We have
ASIR, adaptive statistical iterative reconstruction; CTDI, volume computed tomography dose index; DLP, dose-length product; IRIS, image reconstruction in space.
ASIR ASIR ASIR ASIR IRIS iDose 26.8/— 24/— 21.3/127 —/135.08 19.1/121.2 —/162.5 860 470 306 380 182.8 263 17.0 12.0 6.15 7.7 3.5 5.11 1,193 894 1054 577 365.7 526 25.0 22.0 20.79 12 7.0 11.43 Noise index selection noise index selection Low kVp (80) Noise index selection Low mAs (100) Low mAs (100–120) Sagara (29) Hara (27) Kaza (28) Kambadakone (30) Lee (5) Authors study
Mean Dose Modified DLP (mGy-cm) Mean Dose Modified CTDI (mGy) Mean Previous DLP (mGy-cm) Mean Previous Dose CTDI (mGy) Dose Reduction Technique Author
TABLE 3. Dose Reduction on Dose-modified Abdominal CECT Scans Using Different Reconstruction Techniques
Mean Body Mass Index/Body Weight (lb)
Iterative Reconstruction Technique
Academic Radiology, Vol 20, No 11, November 2013
DOSE-MODIFIED CT SCANS OF THE ABDOMEN USING LOW TUBE CURRENT AND HYBRID ITERATIVE RECONSTRUCTION
demonstrated a 55% dose reduction that does not compromise diagnostic quality when hybrid IRT is used to reconstruct the CT images. Conventional FBP failed to generate diagnostic quality images in 29% of the same DM-CT scan image series. The clinical claims of h-IRT for improving image quality in abdominal CT exams performed at 33% to 70% lower dose than FBP protocols have been reported with other reconstruction algorithms (5,27–30). However, to appreciate dose improvements better and compare radiation dose more accurately between these studies, more quantitative measures such as absolute CTDI or DLP in mGy-cm are appropriate. Our mean CTDI values for patients weighing less and more than 90 kg were 5.0 mGy and 5.5 mGy, respectively, and these doses are at the lower end of the spectrum of abdomen CT dose from other published studies in patients with similar body size (5,27–30) (Table 3). In a recent publication, Nakaura et al reported very encouraging results with the use of hybrid iterative reconstruction in multiphasic liver CT at low kVp settings (13). The authors demonstrated the feasibility of reducing both dose and iodinated contrast medium use by 50% and 40%, respectively, while preserving image quality and contrast-to-noise ratio with hybrid iterative reconstruction in small size adults. Although using low kVp is desirable for its dual benefit of improving image contrast and lowering CT dose, its application in portal venous phase examinations is still evolving (14). Patient body size in North America remains a challenge with low kVp use. This is further compounded by incomplete understanding of appropriate kVp and mA selection for each patient to meet image quality expectations (31). But low kVp use has been more popular for CT examinations with high intrinsic contrast such as arterial phase acquisitions and CT angiography (32–36). Moreover, given the recent introduction of this hybrid reconstruction technique in our practice, we have implemented a graded approach to gain experience, familiarize the radiologists with the image quality, and learn about any potential limitations in interpreting dose-modified CT exams. Indeed, we have discovered that the influence level of h-IRT affects the quality of CT images because DM-CT data reconstructed with h-IRT level 1 were rated as nondiagnostic quality in 12% of the 25 patients to which it was applied. Our phantom study results are similar to observations made by other investigators using the Philips hybrid reconstruction algorithm and other partial iterative reconstruction approaches. iDose allows significant improvement in image noise over FBP regardless of the influence level of hybrid reconstruction applied (5,26,35). Our study has a few limitations. First, the sample size is relatively small to reliably assess the impact of anthropometric variables and different body sizes on image quality. We did not encounter enough patients with very large body habitus. Moreover, our study did not evaluate the utility of this hybrid reconstruction on lesion conspicuity and diagnosis of various abnormalities in the abdomen because we have studied patients with varying clinical indications. However, the main objective of assessing the performance this iterative 1411
FUENTES-ORREGO ET AL
reconstruction technique had on image quality in dosemodified CT scans of the abdomen was achieved. Indeed, we have used image quality and reader confidence to assess radiologist acceptance of these images in rendering diagnostic interpretations. Finally, the possible effect of using low tube potential settings on lowering radiation dose and the impact of h-IRT on image quality was not investigated in this pilot study because we do not practice low-kVp acquisitions for portal venous phase exams. In conclusion, our study shows that hybrid iterative reconstruction technique (iDose4) can be reliably applied to improve image quality and noise in dose-modified abdominal CECT scans. Using our described approach, a 55% drop in the CT radiation dose can be accomplished over our previous dose FBP images, which could be advantageous to all patients, especially benefitting smaller and averaged sized patients.
REFERENCES 1. Boone JM. Multidetector CT opportunities, challenges, and concerns associated with scanners with 64 or more detector rows. Radiology 2006; 241:334–337. 2. Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med 2007; 357:2277–2284. lez A, Mahesh M, Kim K, et al. Projected cancer risks 3. Berrington de Gonza from computed tomographic scans performed in the United States in 2007. Arch Intern Med 2009; 169:2071–2077. 4. Thrall JH. Radiation exposure in CT scanning and risk: where are we? Radiology 2012; 264:325–328. 5. Lee SJ, Park SH, Kim AY, et al. A prospective comparison of standarddose CT enterography and 50% reduced-dose CT enterography with and without noise reduction for evaluating Crohn disease. AJR Am J Roentgenol 2011; 197:50–57. 6. Fleischmann D, Boas FE. Computed tomography—old ideas and new technology. Eur Radiol 2011; 21:510–517. €ller HC, et al. Iterative reconstruction 7. Schindera ST, Diedrichsen L, Mu algorithm for abdominal multidetector CT at different tube voltages: assessment of diagnostic accuracy, image quality, and radiation dose in a phantom study. Radiology 2011; 260:454–462. 8. Marin D, Nelson RC, Schindera ST, et al. Low-tube-voltage, high-tubecurrent multidetector abdominal CT: improved image quality and decreased radiation dose with adaptive statistical iterative reconstruction algorithm—initial clinical experience. Radiology 2010; 254:145–153. 9. Prakash P, Kalra MK, Kambadakone AK, et al. Reducing abdominal CT radiation dose with adaptive statistical iterative reconstruction technique. Invest Radiol 2010; 45:202–210. 10. Pickhardt PJ, Lubner MH, Kim DH, et al. Abdominal CT with model-based iterative reconstruction (MBIR): initial results of a prospective trial comparing ultralow-dose with standard-dose imaging. AJR Am J Roentgenol 2012; 199:1–9. €ger M, Dahlstro € m N, et al. Radiation dose reduction 11. Kalra MK, Woisetschla with sinogram affirmed iterative reconstruction technique for abdominal computed tomography. J Comput Assist Tomogr 2012; 36:339–346. 12. Nakaura T, Awaki K, Maruyama N, et al. Abdominal dynamic CT in patients with renal dysfunction: contrast agent dose reduction with low tube voltage and high tube current-time product settings at 256-detector row CT. Radiology 2011; 261:467–476. 13. Nakaura T, Nakamura S, Maruyama N, et al. Low contrast agent and radiation dose protocol for hepatic dynamic CT of thin adults at 256-detector row CT: effect of low tube voltage and hybrid iterative reconstruction algorithm on image quality. Radiology 2012; 264:445–454. ~es LS, Fletcher JG, Harmsen WS, et al. Appropriate patient selec14. Guimara tion at abdominal dual-energy CT using 80 kV: relationship between patient size, image noise, and image quality. Radiology 2010; 257:732–742. 15. Zalis ME, Perumpillichira JJ, Kim JY, et al. Polyp Size at CT colonography after electronic subtraction cleansing in an anthropomorphic colon phantom. Radiology 2005; 236:118–124.
1412
Academic Radiology, Vol 20, No 11, November 2013
16. European Commission European guidelines on quality criteria for computed tomography. (EUR 16262 EN). Luxembourg: European Commission & The Office For Official Publications of the European Communities, 2000. www. drs.dk/guidelines/ct/quality/Download.htm. Accessed July 10, 2013. 17. Singh S, Kalra MK, Hsieh J, et al. Abdominal CT: comparison of adaptive statistical iterative and filtered back projection reconstruction techniques. Radiology 2010; 257:373–383. 18. American Association of Physicists in Medicine. Size-specific dose estimates (SSDE) in pediatric and adult body CT examinations. Task Group 204. College Park, MD: American Association of Physicists in Medicine, 2011. 19. Tack D, Bohy P, Perlot I, et al. Suspected acute colon diverticulitis: imaging with low-dose unenhanced multi-detector row CT. Radiology 2005; 237: 189–196. 20. Lee SH, Kim MJ, Yoon CS, et al. Radiation dose reduction with the adaptive statistical iterative reconstruction (ASIR) technique for chest CT in children: an intra-individual comparison. Eur J Radiol 2012; 81:e938–e943, http:// www.sciencedirect.com/science/article/pii/S0720048X1200280X. 21. Liu W, Esler SJ, Kenny BJ, et al. Low-dose non enhanced helical CT of renal colic: assessment of ureteric stone detection and measurement of effective dose equivalent. Radiology 2000; 215:51–54. 22. Marin D, Nelson RC, Barnhart H, et al. Detection of pancreatic tumors, image quality, and radiation dose during the pancreatic parenchymal phase: effect of a low-tube-voltage, high-tube-current CT technique—preliminary results. Radiology 2010; 256:450–459. 23. Kulkarni NM, Uppot RN, Eisner BH, et al. Radiation dose reduction at multidetector CT with adaptive statistical iterative reconstruction for evaluation of urolithiasis: how low can we go? Radiology 2012; 265:158–166. 24. Baker ME, Dong F, Primak A, et al. Contrast-to-noise ratio and lowcontrast object resolution on full- and low-dose MDCT: SAFIRE versus filtered back projection in a low-contrast object phantom and in the liver. AJR Am J Roentgenol 2012; 199:8–18. 25. Negi N, Yoshikawa T, Ohno Y, et al. Hepatic CT perfusion measurements: A feasibility study for radiation dose reduction using new image reconstruction method. Eur J Radiol 2012; 81:3048–3054. €l PB, Fingerle AA, Renger B, et al. Initial performance characterization 26. Noe of a clinical noise-suppressing reconstruction algorithm for MDCT. AJR Am J Roentgenol 2011; 197:1404–1409. 27. Hara AK, Paden RG, Silva AC, et al. Iterative reconstruction technique for reducing body radiation dose at CT: feasibility study. AJR Am J Roentgenol 2009; 193:764–771. 28. Kaza RK, Platt JF, Al-Hawary MH, et al. CT enterography at 80 kVp with adaptive statistical iterative reconstruction versus at 120 kVp with standard reconstruction: image quality, diagnostic adequacy and dose reduction. AJR Am J Roentgenol 2012; 198:1084–1092. 29. Sagara Y, Hara A, Pavlicek W, et al. Abdominal CT: comparison of low dose CT with adaptive statistical iterative reconstruction and routine dose CT with filtered back projection in 53 patients. AJR Am J Roentgenol 2010; 195:713–719. 30. Kambadakone AR, Chaudhary NA, Desai GS, et al. Low dose MDCT and CT enterography of patients with Crohn disease: feasibility of adaptive statistical iterative reconstruction. AJR Am J Roentgenol 2011; 196: W743–W752. 31. Yu L, Li H, Fletcher JG, et al. Automatic selection of tube potential for radiation dose reduction in CT: a general strategy. Med Phys 2010; 37: 234–243. 32. Iezzi R, Cotroneo AR, Giammarino A, et al. Low-dose multidetector-row CT-angiography of abdominal aortic aneurysm after endovascular repair. Eur J Radiol 2011; 79:21–28. 33. Sahani DV, Kalva SP, Hahn PF, et al. 16-MDCT angiography in living kidney donors at various tube potentials: impact on image quality and radiation dose. AJR Am J Roentgenol 2007; 188:115–120. 34. Namimoto T, Oda S, Utsunomiya D, et al. Improvement of image quality at low-radiation dose and low-contrast material dose abdominal CT in patients with cirrhosis: intraindividual comparison of low tube voltage with iterative reconstruction algorithm and standard tube voltage. J Comput Assist Tomogr 2012; 36:495–501. 35. Kalva SP, Sahani DV, Hahn PF, et al. Using the K-edge to improve contrast conspicuity and to lower radiation dose with a 16-MDCT: a phantom and human study. J Comput Assist Tomogr 2006; 30:391–397. 36. Pinho DF, Kulkarni NM, Krishnaraj A, et al. Initial experience with singlesource dual-energy CT abdominal angiography and comparison with single-energy CT angiography: image quality, enhancement, diagnosis and radiation dose. Eur J Radiol 2013; 23:351–359.