Materials and Methods From April 2009 until February 2010, fifteen potential donors for living-related liver transplantation (mean age 37.6 ± 10.8 years) were included. Potential donors underwent a two-phase CT-examination of the liver. The first phase consisted of a single-energy non-enhanced CT-acquisition that provided TNI. After administration of hepatobiliary contrast agent, the second phase was performed as a dual-energy cholangiographic CT-acquisition. This provided VNI. Objective image quality (attenuation values [bile ducts and liver parenchyma] and contrast-to-noise ratio) and subjective overall image quality (1 – excellent; 5 – non diagnostic) were evaluated. Effective radiation dose was compared. Results For TNI and VNI, attenuation values for bile ducts were 16.8 ± 11.2 HU and 5.5 ± 17.0 HU ( p < 0.05) and for liver parenchyma 55.3 ± 8.4 HU and 58.1 ± 10.6 HU (n.s.). For TNI and VNI, contrast-to-noise ratio was 2.6 ± 0.6 HU and 6.9 ± 2.1 HU ( p < 0.001). For VNI, subjective overall image quality was 1 in ten datasets, 2 in four datasets and 3 in one dataset. Effective radiation dose for the dual-energy cholangiographic CT-acquisition was 3.6 ± 0.9mSv and for two-phase single-energy CT-cholangiography 5.1 ± 1.3mSv ( p < 0.001). Conclusion In this study on iodine removal in intravenous dual-energy CT-cholangiography, subjective image quality is equivalent, contrast-to-noise ratio is improved and effective radiation dose is reduced when VNI is performed. The differences between TNI and VNI with respect to attenuation values seem to have limited clinical relevance and therefore we consider VNI as effective to replace TNI. Keywords CT Dual-energy Iodine removal Virtual non-enhanced images Cholangiography 1 Introduction According to the clinical indication, standard CT-examinations of the abdomen include non-enhanced and contrast-enhanced acquisitions in different phases (e.g. arterial and porto-venous) [1] . Non-enhanced images are important to identify calcifications and fat [2] . They are the basis for quantification of contrast-uptake in abdominal masses. Non-enhanced acquisitions are important to delineate artificial materials (e.g. clips and drainages). They are essential in patients suspected of having active hemorrhage for the identification of the bleeding site [3] . CT-cholangiography with administration of a hepatobiliary contrast agent opacifies the biliary system [4] . Comparable to abdominal CT-examinations with administration of an extracellular contrast agent, non-enhanced images before infusion of a hepatobiliary contrast agent are essential to make correct diagnoses of the biliary system (e.g. stones, leaks, hemobilia, and tumors) [5,6] . Modern CT-systems provide the possibility of dual-energy imaging. Independent of the technical realization (either as dual-source scanners with two X-ray tubes and detectors or by rapid switching of the tube voltage of a single-tube scanner), a dual-energy CT-acquisition is performed as acquisition of two different raw datasets [2] . The doubled photon spectra information allows superb differentiation of iodine due to its high atomic number [7] . Dual-energy post-processing on the basis of the three-material decomposition theory (soft tissue versus fat versus iodine) can remove iodine from contrast-enhanced images resulting in virtual non-enhanced images [8] . Virtual non-enhanced imaging (VNI) derived from contrast-enhanced dual-energy CT-acquisitions has proven efficacy in multiple clinical investigations [2,9,10] . However, all published data covers extracellular contrast agents. In contrast, our study evaluates the feasibility of VNI after administration of an intracellular, hepatobiliary contrast agent. The purpose was to evaluate whether VNI is effective to replace true non-enhanced imaging (TNI) applying iodine removal in intravenous dual-energy CT-cholangiography. 2 Materials and methods 2.1 Patients From April 2009 until February 2010, dual-energy CT image data of fifteen potential donors for living-related liver transplantation (nine women and six men) were acquired. Main clinical data included a mean age of 34.4 ± 9.6 years (23–48 years), a mean weight of 73.4 ± 9.1 kg (range 50.1–84.0 kg), a mean height of 176.2 ± 7.3 cm (range 169–189 cm), a mean body mass index (BMI) of 25.3 ± 3.3 kg/m 2 (range 22.4–29.8 kg/m 2 ), a mean bilirubin level of 0.7 ± 0.2 mg/dl (range 0.3–1.3 mg/dl), a mean pseudocholine esterase level of 9.1 ± 2.2 U/l (range 5.3–13.9 U/l), a mean albumine level of 44.7 ± 3.2 g/l (range 39.1–47.8 g/l) and a mean thromboplastin time of 99.6 ± 12.1 s (range 72.0–119.6 s). None of the potential donors had a history of adverse reactions to iodinated contrast agents, renal insufficiency (creatinine >1.5 mg/dl), elevated serum bilirubin levels (>1.5 mg/dl) or hepatobiliary surgery. All potential donors underwent preoperative evaluation of the biliary system applying intravenous dual-energy CT-cholangiography for which they gave written informed consent. In our institution, CT-cholangiography constitutes one of the final diagnostic steps in the evaluation of potential donors for living-related liver transplantation. The prospective study at hand was approved by the Institutional Review Board and compliant with the requirements of the Health Insurance Portability and Accountability Act. 2.2 CT-examination All potential donors received standard premedication with 400 mg cimetidine (Ranitic ® , Hexal, Holzkirchen, Germany) and 4 mg clemastinhydrogenfumarat (Tavegil ® , Novartis, Munich, Germany) to prevent adverse contrast agent reactions. The examinations were performed with the same 64-slice multi-row detector dual-source CT-scanner (Somatom Definition DS, Siemens Medical Solutions, Forchheim, Germany). The liver was studied with a two-phase protocol. The first phase consisted of a single-energy non-enhanced CT-acquisition (64 × 0.6 mm collimation; 120 kVp tube potential; quality reference tube current of 240 mAs; 0.5 s gantry rotation time; pitch 0.55) providing TNI. Subsequently, we performed slow and continuous infusion of the hepatobiliary contrast agent (100 ml of meglumine iotroxate, Biliscopin ® ; Bayer Schering, Berlin, Germany) using an automated infusion pump with an injection rate of 150 ml/h [11] . Five minutes after completion of the contrast agent infusion, the second phase was performed as a dual-energy cholangiographic CT-acquisition (64 × 0.6 mm collimation; 140 kVp and 80 kVp tube potential; quality reference tube current of 96 mAs (140 kVp) and 404 mAs (80 kVp); 0.5 s gantry rotation time; pitch 0.55) [5] . The potential donors were off-center positioned in the gantry to center the liver within the smaller field-of-measurement of the second detector (Ø FOV(B) = 26.0 cm). The scan lengths of the single-energy non-enhanced CT-acquisition and of the dual-energy cholangiographic CT-acquisition were selected to be identical. The dual-energy dataset provided VNI and true enhanced imaging (TEI). 2.3 Image reconstruction For abdominal dual-energy CT, recommended collimation is 14 mm × 1.2 mm [12] . In our study, we selected a thinner collimation to reduce partial volume effects and to increase spatial resolution as much as possible in order to increase the visualization of the tiny high-order bile ducts. For the single-energy non-enhanced CT-acquisition and for the dual-energy cholangiographic CT-acquisition, overlapping axial images of 0.75 mm slice thickness were reconstructed with an increment of 0.4 mm. Further image parameters included a soft tissue window with a center of 56 HU and a width of 342 HU. Medium soft body kernels were used for image reconstructions of the single-energy non-enhanced CT-acquisitions (B20f) and dual-energy cholangiographic CT-acquisitions (B20 and D20). TEI were reconstructed using both raw datasets with the recommended weighting ratio (30% 140 kVp, 70% 80 kVp) [2] . VNI was performed on the basis of the dual-energy application (“Liver-VNC ® ”, Siemens Medical Solutions, Forchheim, Germany) [2] . To obtain VNI from the dual-energy cholangiographic CT-acquisition, the basic presets included an iodine image overlay of 0% (yielding “no” iodine images). 2.4 Image analysis – general considerations Axial images were evaluated independently by two different pairs of radiologists (one and six years experience of abdominal imaging as well as four and ten years experience of abdominal imaging) on a picture archiving and communication system work station (GE Centricity 4.1, GE Healthcare, Barrington, USA). TNI, VNI and TEI were evaluated side-by-side. The information of the TEI was important for definition of the bile ducts and the renal pelvic system. Image analysis included an objective and a subjective part. Objective image analysis was performed blinded to the subjective image analysis by the identical pairs of radiologists. For objective image analysis, regions of interest (ROIs) in TNI, VNI and TEI had to be placed in corresponding positions. Since the z-positions of the anatomic structures were different in TNI and VNI/TEI, manual correction after co-registration was necessary. Therefore, anatomical landmarks (e.g. portal vein, hilar fat, liver veins and hepatic ligaments) were used. 2.5 Objective image quality 2.5.1 Attenuation values For TNI, VNI and TEI, attenuation values (in Hounsfield units [HU]) were determined for bile ducts, gallbladder, liver, kidney, spleen, psoas muscle and aorta [2,9] . Size, shape and position of the ROIs were kept constant between the three different datasets. For bile ducts, the attenuation values were calculated as mean of four attenuation measurements (one in the right and one in the left intrahepatic main duct as well as two in the common duct) in circular ROIs with an area of 4–6 mm 2 . For gallbladder, liver, kidney, spleen, psoas muscle and aorta, the attenuation values were calculated as mean values of three attenuation measurements in circular ROIs with an area of 1–3 cm 2 . 2.5.2 Image noise, signal-to-noise ratio and contrast-to-noise ratio For TNI, VNI and TEI, image-noise was defined for bile ducts, gallbladder, liver, kidney, spleen, psoas muscle and aorta as the mean standard deviation (SD) of the particular attenuation measurements. For TNI, VNI and TEI, duct-to-liver signal-to-noise ratio (SNR) and duct-to-liver contrast-to-noise ratio (CNR) were calculated [5] : (1) SNR = | ROI duct | S D liver (2) CNR = | ( ROI duct - ROI liver ) | S D liver where ROI duct is the bile duct attenuation, ROI liver is the liver attenuation, and SD liver is the image noise in the liver. For VNI, the image quality score for exclusion of any liver parenchyma or extrahepatic bile ducts by the limited field-of-view of the 26 cm detector (on a scale of 0 to 4: 0 – none; 1 – ≤1 cm; 2 – ≤2 cm; 3 – ≤3 cm; and 4 – >3 cm) was determined. 2.6 Subjective image quality As specified by Graser et al., three different scores for the subjective image quality of VNI were applied [2] . First, the image quality for artifacts (on a scale of 0–3: 0 – none; 1 – mild; 2 – moderate; and 3 – severe) was determined. Second, the image quality for overall image quality (on a scale of 1–5: 1 – excellent; 2 – good; 3 – fair; 4 – poor; and 5 – non diagnostic) was analyzed. Third, the image quality for VNI as replacement for TNI (on a scale of 1–3: 1 – complete; 2 – with restrictions; and 3 – not acceptable) was evaluated. 2.7 Effective radiation dose Effective radiation doses were analyzed for the dual-energy cholangiographic CT-acquisition (ED DE ) and for two-phase single-energy CT-cholangiography (ED TP ). ED DE was calculated from the dose-length-product (DLP) of the dual-energy acquisition as protocoled in the scan summary (DLP DE ), using the conversion factor for abdominal CT-acquisitions ( c = 0.015 mSv/mGycm) [5] : (3) E D DE = c × DLP DE ED TP is the sum of effective doses of (1) the single-energy non-enhanced CT-acquisition (ED SE ) and (2) a “virtual” single-energy cholangiographic CT-acquisition (ED SE_virtual ): (4) E D TP = E D SE + E D SE _ virtual ED SE was calculated in analogy to Eq. (3) . ED SE_virtual was calculated on the basis of the scan summary value of DLP for the dual-energy cholangiographic CT-acquisition. To this end, a method previously introduced by our group was used (Schwarzwaelder et al., RSNA 2010, SSC14-08). Based on phantom measurements of CTDI at tube voltages of 80 kVp, 120 kVp and 140 kVp, this method permits calculation of the radiation dose for a conventional 120 kVp CT-acquisition equivalent to the dual-energy CT-acquisition conducted in reality, using the comprehensive DLP value listed in the scan summary for this dual-energy CT-acquisition. In other words, DLP DE was mathematically rescaled in such a manner that its resulting value corresponds to an identical acquisition performed with a single tube-detector assembly at 120kVp. This resulted in the DLP of the “virtual” single-energy cholangiographic CT-acquisition (DLP SE_virtual ). Using this value, ED SE_virtual was calculated in analogy to equation (3) : (5) E D SE _ virtual = c × DLP SE _ virtual These considerations permit a fair comparison of the radiation doses of the dual-energy cholangiographic CT-acquisition (providing VNI and TEI) and two-phase single-energy CT-cholangiography (providing TNI and TEI both generated with single-energy technique) under standardized conditions. 2.8 Statistics For attenuation values, image noise, SNR and CNR, descriptive and comparative statistics were performed. Since the results of both pairs of radiologists were comparable, their data was averaged. For the effective radiation dose, descriptive and comparative statistics were performed. The numerical data is presented as mean with SD and range. TNI and VNI were compared applying the non-parametric Wilcoxon signed rank test. P < 0.05 was considered to be statistically significant. The statistical analysis was carried out with SAS software (Version 9.1, SAS Institute Inc., Cary, NC, USA). For the image quality score for exclusion of any liver parenchyma or extrahepatic bile ducts by the limited field-of-view of the 26 cm detector as well as for the three different scores for the evaluation of subjective image quality of VNI, descriptive statistics was performed. For the latter, the individual data of both pairs of radiologists is presented. 3 Results All examinations have been completed as planned. Adverse effects after infusion of the hepatobiliary contrast agent were not observed. All potential donors left our department in excellent clinical condition. In the end, six patients underwent liver donation. 3.1 Objective image quality 3.1.1 Attenuation values For TNI and VNI, mean attenuation values were significantly different for bile ducts (16.8 ± 11.2 HU versus 5.5 ± 17.0 HU; p < 0.05), for the gallbladder (14.3 ± 7.2 HU versus −1.8 ± 13.1 HU; p < 0.001), for the kidney (29.5 ± 3.5 HU versus 23.8 ± 3.5 HU; p < 0.001) and for the spleen (43.1 ± 6.4 HU versus 51.6 ± 7.2 HU; p < 0.001) ( Table 1 ) ( Figs. 1 and 2 ). For the liver, psoas muscle and aorta, mean attenuation values for TNI and VNI did not differ significantly. 3.1.2 Image noise, signal-to-noise ratio and contrast-to-noise ratio For TNI and VNI, mean image noise was significantly different for all evaluated structures, e.g. for the bile ducts (14.8 ± 3.2 HU versus 6.5 ± 2.2 HU; p < 0.001), for the gallbladder (13.6 ± 2.7 HU versus 7.4 ± 1.5 HU; p < 0.001) and for the liver (14.8 ± 2.0 HU versus 7.6 ± 0.6 HU; p < 0.001) ( Table 2 ) ( Fig. 1 ). For VNI, mean image noise was always about half that of TNI. For TNI and VNI, mean SNR was not significantly different (1.2 ± 0.9 versus 0.7 ± 2.2; n.s.) ( Table 3 ). For TNI and VNI, mean CNR was significantly different (2.6 ± 0.6 versus 6.9 ± 2.1; p < 0.001). For VNI, image quality for exclusion of any liver parenchyma and extrahepatic bile ducts with the 26 cm field-of-view detector was 0 in eight datasets, 1 in three datasets, 2 in two datasets and 3 in two datasets ( Table 4 ) ( Fig. 2 ). 3.2 Subjective image quality For VNI, image quality for artifacts was 0 in five datasets, 1 in four datasets and 2 in six datasets ( Table 4 ). Nine datasets demonstrated iodine subtraction artifacts and five datasets ring artifacts ( Fig. 3 ). One dataset had subtraction and ring artifacts. For VNI, image quality for overall image quality was 1 in ten datasets, 2 in four datasets and 3 in one dataset. For VNI, image quality for VNI as a replacement for TNI was 1 in all datasets. 3.3 Effective radiation dose The mean effective radiation dose for the dual-energy cholangiographic CT-acquisition was 3.6 ± 0.9 mSv (2.1–4.9 mSv) and for two-phase single-energy CT-cholangiography 5.1 ± 1.3 mSv (3.4–7.8 mSv) ( p < 0.001). In consequence, the dual-energy cholangiographic CT-acquisition resulted in a mean dose reduction of 28.4 ± 12.6% (11.0–57.6%). 4 Discussion In this study, we evaluated whether VNI is effective to replace TNI applying iodine removal in intravenous dual-energy CT-cholangiography. Our results demonstrate that subjective image quality is comparable between both datasets. From a quantitative point of view, SNR is equivalent and CNR is improved in VNI compared to TNI. Moreover, the effective radiation dose is reduced when VNI is performed. However, differences between both datasets exist with respect to attenuation values. 4.1 Objective image quality 4.1.1 Attenuation values As in other publications, attenuation values for liver and muscle were comparable between VNI and TNI [2,9] . In contrast, our study detected significantly different attenuation values for kidney and spleen. While Zhang et al. published different attenuation values for the aorta for VNI and TNI, this could neither be confirmed by Graser et al. nor by our group [2] . Regarding the biliary system, different attenuation values were detected for VNI and TNI. These inconsistent results might be explained with the different classes of intravenous contrast agents (extracellular versus intracellular) as well as with the different phases of the dual-energy acquisitions (arterial, porto-venous or nephrographic phase). Nevertheless, valuable parallels can be noted. For high attenuating tubular structures (e.g. the aorta after injection of iopromid [Ultravist ® , Bayer Schering, Berlin Germany] or the biliary system after administration of meglumine iotroxate), the post-processing algorithm for VNI seems to have limitations. Optimal iodine subtraction seems to be not feasible to the extent that VNI is identical with TNI. This observation is in line with another publication reporting attenuation values for the aorta of 34.9 HU for VNI compared to 40.9 HU for TNI [9] . We confirm these results methodologically, since the attenuation values for bile ducts and gallbladder differed significantly for VNI and TNI. A similar trend might also exist for intensely enhancing parenchymal organs. In their paper on VNI reconstructed from a dual-energy nephrographic acquisition, Graser et al. mentioned that the high range of attenuation values for VNI “reflects the fact, that dual-energy based iodine subtraction is not always perfect” [2] . 4.1.2 Image noise, signal-to-noise ratio and contrast-to-noise ratio Our study detected higher image noise for TNI compared to VNI. Other publications indicate similar results [2,9] . The smaller range of image noise in VNI indicates a more homogeneous overall image quality. Another interesting finding of this study is that image noise for VNI is approximately half as much as for TNI. As a rule of thumb, the reduction of image noise by 50% requires a quadruplication of the radiation dose. When compared to TNI, the image impression of our VNI is as if the corresponding CT-acquisition had been performed with quadrupled radiation dose. This is even more stunning since VNI results from subtraction of two image datasets, what normally leads to an addition of the image noise of both datasets. Consequently, noise in VNI should be about twice as high as for images resulting from TNI (even though radiation dose for the dual-energy cholangiographic CT-acquisitions was approximately 35% higher compared to the single-energy non-enhanced CT-acquisitions). It is very likely that the VNI post-processing application involves relevant “artificial voxel smoothing”. Or in other words, the absolute difference of attenuation between adjacent voxels is lower in VNI compared to TNI. This results in a “washy” image impression of VNI. In our study, SNR was equivalent in VNI and TNI, which is different to another publication [9] . Theoretically, image quality should increase with specific dual-energy post-processing. Increased SNR could be demonstrated in pure iodine images and optimized-contrast images, but not in images post-processed without dual-energy techniques [5] . This effect might be also due to the use of a special “D” kernel improving material characterization and reducing blurring [9] . The available data on CNR is confusing. Zhang et al. published identical CNR for VNI and TNI [9] . In a study with evaluation of CNR between intracranial hematoma and brain tissue, Ferda et al. found clearly lower CNR in VNI [13] . In this context, it is important to mention that dual-energy iodine subtraction is based on the increased X-ray attenuation of iodine at 80 kV compared to 140 kV [8] . The high density of intracranial hematomas is predominantly caused by iron as an element of clot-forming hemoglobin. As a consequence, comparisons between structures containing iron and iodine lack and CNRs of VNI and TNI are not comparable. We detected significantly higher CNR in VNI. From a more practical point of view, the results of our study demonstrate, that eccentric patient positioning in the CT-gantry allows complete or almost complete coverage of the liver parenchyma and the extrahepatic bile ducts by the 26 cm field-of-view detector. With second-generation scanners, we can assume that dual-energy liver imaging is feasible for obese patients and big livers. 4.2 Subjective image quality Subjective image quality of VNI was diagnostic in all of our subjects. Overall image quality was comparable to other investigations [2,9] . No relevant liver or bile duct anatomy was excluded by the 26 cm field-of-view detector, which can be mostly attributed to proper eccentric patient positioning in the CT-gantry. Artifacts, although existent in ten datasets, were mild or moderate. The vast majority of artifacts were iodine subtraction artifacts, primarily resulting from an intensely opacified gallbladder. These artifacts were maximal variants of the imperfect VNI post-processing algorithm [2] . In contrast to the publication of Zhang et al., ring artifacts in the center of VNI (and TEI) were not so common in our datasets [9] . Optimized configuration of the CT-system (e.g. dual-energy scanning in multiple patients in a row or directly after system check-up) might further reduce ring artifacts. 4.3 Effective radiation dose Zhang et al. detected a significantly lower DLP for dual-phase CT (consisting of a dual-energy arterial acquisition and a dual-energy porto-venous acquisition; VNI reconstructed from contrast-enhanced acquisitions) compared to triple-phase CT (consisting of a true non-enhanced acquisition, a dual-energy arterial acquisition and a dual-energy porto-venous acquisition) [9] . Graser et al. published a significantly decreased effective radiation dose for a dual-phase protocol (consisting of a dual-energy nephrographic and a delayed acquisition; VNI reconstructed from contrast-enhanced acquisitions) compared to a triple-phase protocol (consisting of a true non-enhanced acquisition, a dual-energy nephrographic and an additional delayed acquisition) [2] . However, these considerations are too simple. First, it is trivial that a CT-protocol results in lower radiation exposure when one phase is omitted but all other parameters stay the same. Second, a perfect dual-energy CT-examination would not include both a true non-enhanced and a dual-energy acquisition, since non-enhanced images can be calculated from the dual-energy acquisition alone. For dose calculations, a CT-protocol should include either a true non-enhanced acquisition followed by single-energy acquisitions or one could save the true non-enhanced acquisition and involve a dual-energy acquisition with VNI. In this manuscript, we calculated the effective radiation dose of a “virtual” single-energy cholangiographic CT-acquisition from 1. the scan summary value of DLP for the dual-energy cholangiographic CT-acquisition and 2. phantom measurements of CTDI at tube voltages of 80 kVp, 120 kVp and 140 kVp. The DLP of the dual-energy cholangiographic CT-acquisition was mathematically rescaled in such a manner that its resulting value corresponds to an identical acquisition acquired with a single tube-detector assembly at 120 kVp. In the end, our calculations yield an effective dose reduction of approximately 30% when the dual-energy cholangiographic CT-acquisition was compared to two-phase single-energy CT-cholangiography. The dose-saving potential of dual-energy CT-cholangiography is another reason to implement this technique as a diagnostic alternative in the clinical routine [5] . In doing so, the differences between TNI and VNI with respect to attenuation values have to be kept in mind. 4.4 Study limitations There are limitations to this study. Only a small number of healthy subjects were examined. The reduced field-of-view of the second detector might compromise hepatobiliary imaging in obese patients and for big livers. Postoperative states with bile duct prostheses, sutures and clips can increase artifacts and reduce the diagnostic value of VNI. Finally, the radiation dose calculations based on knowledge acquired from phantom measurements. Although the dose values were calculated (but not directly measured) for the “virtual” single-energy cholangiographic CT-acquisition, this procedure seems to be a valid method to convert the comprehensive radiation dose of a dual-energy CT-acquisition into the radiation dose value of a comparable single-energy CT-acquisition (at a tube potential of 120 kVp). Thus, one is able to calculate the dose value of a single-energy CT-acquisition with the same scan protocol settings (apart from tube potential and rescaled current) as for the dual-energy CT-acquisition. This allows for direct and fair effective dose considerations of the different CT-techniques (single-energy versus dual-energy). 5 Conclusion In this study on iodine removal in intravenous dual-energy CT-cholangiography, subjective image quality is equivalent, CNR is improved and effective radiation dose is reduced when VNI is performed. The differences between TNI and VNI with respect to attenuation values seem to have limited clinical relevance and therefore we consider VNI as effective to replace TNI. Conflict of Interest The corresponding author declares, that no conflict of interest exists for all involved authors. Acknowledgement This study was supported by Siemens Medical Solutions, Forchheim, Germany . References [1] D.H. Szolar F. Kammerhuber S. Altziebler Multiphasic helical CT of the kidney: increased conspicuity for detection and characterization of small (< 3-cm) renal masses Radiology 202 1997 211 217 [2] A. Graser T.R. Johnson E.M. Hecht Dual-energy CT in patients suspected of having renal masses: can virtual nonenhanced images replace true nonenhanced images? Radiology 252 2009 433 440 [3] C.M. Sommer U. Stampfl N. Bellemann Patients with life-threatening arterial renal hemorrhage: CT angiography and catheter angiography with subsequent superselective embolization Cardiovascular and Interventional Radiology 33 2010 498 508 [4] R.S. Breiman F.V. Coakley E.M. Webb CT cholangiography in potential liver donors: effect of premedication with intravenous morphine on biliary caliber and visualization Radiology 247 2008 733 737 [5] C.M. Sommer C.B. Schwarzwaelder W. Stiller Dual-energy computed-tomography cholangiography in potential donors for living-related liver transplantation: initial experience Investigative Radiology 45 2010 406 412 [6] R.N. Gibson J.M. Vincent T. Speer N.A. Collier K. Noack Accuracy of computed tomographic intravenous cholangiography (CT-IVC) with iotroxate in the detection of choledocholithiasis European Radiology 15 2005 1634 1642 [7] M. Petersilka H. Bruder B. Krauss K. Stierstorfer T.G. Flohr Technical principles of dual source CT European Journal of Radiology 68 2008 362 368 [8] T.R. Johnson B. Krauss M. Sedlmair Material differentiation by dual energy CT: initial experience European Radiology 17 2007 1510 1517 [9] L.J. Zhang J. Peng S.Y. Wu Liver virtual non-enhanced CT with dual-source, dual-energy CT: a preliminary study European Radiology 20 2010 2257 2264 [10] N. Takahashi T.J. Vrtiska A. Kawashima Detectability of urinary stones on virtual nonenhanced images generated at pyelographic-phase dual-energy CT Radiology 256 2010 184 190 [11] M. Okada J. Fukada K. Toya R. Ito T. Ohashi A. Yorozu The value of drip infusion cholangiography using multidetector-row helical CT in patients with choledocholithiasis European Radiology 15 2005 2140 2145 [12] H. Chandarana M.C. Godoy I. Vlahos Abdominal aorta: evaluation with dual-source dual-energy multidetector CT after endovascular repair of aneurysms--initial observations Radiology 249 2008 692 700 [13] J. Ferda M. Novak H. Mirka The assessment of intracranial bleeding with virtual unenhanced imaging by means of dual-energy CT angiography European Radiology 19 2009 2518 2522