Automatic Lumen Segmentation in Calcified Plaques: Dual-Energy CT Versus Standard Reconstructions in Comparison With Digital Subtraction Angiography

Transcription

1 Neuroradiology/Head and Neck Imaging Original Research Thomas et al. Neuroradiology/Head and Neck Imaging Original Research Christoph Thomas 1 Andreas Korn 2 Dominik Ketelsen 1 Soeren Danz 2 Ilias Tsifikas 1 Claus D. Claussen 1 Ulrike Ernemann 2 Martin Heuschmid 1 Thomas C, Korn A, Ketelsen D, et al. Keywords: bone removal, CT angiography, dual-energy CT, internal carotid artery, plaque removal DOI: /AJR Received August 29, 2009; accepted after revision November 25, Department of Diagnostic and Interventional Radiology, University of Tübingen, Hoppe-Seyler-Straße 3, Tübingen 72076, Germany. Address correspondence to C. Thomas (Christoph.thomas@med.uni-tuebingen.de). 2 Department of Diagnostic and Interventional Neuroradiology, University of Tübingen, Tübingen, Germany. AJR 2010; 194: X/10/ American Roentgen Ray Society Automatic Lumen Segmentation in Calcified Plaques: Dual-Energy CT Versus Standard Reconstructions in Comparison With Digital Subtraction Angiography OBJECTIVE. Dual-energy CT has the potential to automatically remove calcified plaques from angiographic data sets. The objective of this study is to compare the accuracy of visual grading of stenoses after plaque removal with visual grading in standard reconstructions. Digital subtraction angiography (DSA) was used as a reference standard. SUBJECTS AND METHODS. Twenty-five patients underwent dual-energy CT (140 kv and 80 mas; 80 kv and 234 mas) angiography and DSA. Plaque and bone removal was performed. Twenty-nine calcified stenoses were quantified using standard reconstructions, plaque and bone removal maximum intensity projections after plaque and bone removal, and DSA images, according to the North American Symptomatic Carotid Endarterectomy Trial criteria. The accuracy of the detection of relevant stenoses (> 70%) and occlusions was assessed. Correlation coefficients of the grades of stenoses with DSA were calculated. The influence of vessel enhancement on the accuracy of plaque removal was analyzed. RESULTS. The average postprocessing time was 45 seconds. After plaque removal, all 25 relevant and four nonrelevant stenoses were correctly detected. Six relevant stenoses were overestimated as complete occlusions. With the standard reconstructions, two nonrelevant stenoses were overestimated as relevant. Correlation coefficients (r 2 ) for the grading of stenoses after plaque removal and with standard reconstructions versus DSA were and , respectively. Vessel contrast enhancement correlated weakly (r 2 = ) with the accuracy of plaque removal. CONCLUSION. Dual-energy CT with plaque removal automatically delivers CT luminograms with a high sensitivity for the detection of relevant stenoses and a higher correlation to DSA than standard reconstructions but frequently leads to an overestimation of high-grade stenoses as occlusions. Thus, dual-energy CT plaque and bone removal should be used complementary to standard reconstructions, and not exclusively. D igital subtraction angiography (DSA) is considered the reference standard for diagnostic angiography [1]. No other current technique combines the spatial resolution, contrast, and ability to create luminograms as does DSA. As an alternative noninvasive technique, MRI is limited by high costs and contraindications, such as claustrophobia, cardiac pacemakers, and limited spatial resolution [2]. Because of its wide availability, uncomplicated nature, and noninvasive character, CT is an established and widely accepted technique for imaging the carotid arteries [3]. However, CT is hampered by the depiction of bones and calcified plaques, which makes the creation of luminograms difficult in cases of calcified stenoses and, thus, complicates the quantification of stenoses [4]. Dual-energy CT has the theoretic potential to differentiate between calcified plaques and iodinated contrast medium by exploiting the energy dependence of the CT attenuation of calcium and iodine and might therefore provide luminograms comparable to those generated by DSA [5, 6]. Previous studies have compared the efficiency of bone and plaque removal in dualenergy CT angiography of the supraaortic vessels with that of conventional CT angiography and have found advantages in dual-energy CT in terms of reading time and radiation dose [6 8]. Other studies have reported the feasibility of bone and plaque removal in intracranial and cervical dual-energy CT angiography, with DSA as a reference standard [9, 10]. These studies included unselected patient series with mainly low-to-moderate 1590 AJR:194, June 2010

2 grades of stenosis. In this study, we prospectively examine the accuracy of the grading of stenoses after dual-energy bone and plaque removal in a patient group with known calcified carotid artery stenoses and compare it with visual grading of stenoses in standard reconstructions. Subjects and Methods Patients The local ethics committee agreed to this study. All patients gave their informed consent to participate. Exclusion criteria were known allergic reaction to iodinated contrast medium, impaired renal function, hyperthyroidism, pregnancy, unstable clinical conditions, and the absence of calcified plaques. Twenty-five consecutive patients with known stenoses of the internal carotid arteries who were scheduled for DSA and carotid artery stent angioplasty were examined. Two patients were excluded from the evaluation because they had noncalcified plaques. Thus 23 patients (16 men and seven women) with 29 calcified stenoses (including three bilateral and three tandem stenoses) were prospectively included into the evaluation. The mean age of the included patients was 69 years (range, years). Eight patients were symptomatic. Dual-Energy CT Angiography All examinations were performed with a dualsource CT system (Somatom Definition, Siemens Healthcare). After planning the scan range from the aortic arch to the top of the skull, 70 ml of iodinated contrast medium (400 mg I/mL iomeprol [Imeron 400, Bracco]) at a flow setting of 5.0 ml/s and a 40 ml saline chaser were injected into an antecubital vein with a dual-head injector (CT Stellant, Medrad). Bolus tracking was performed in the aortic arch after a delay of 7 seconds. As soon as contrast medium was visually detected in the aortic arch, the acquisition was triggered manually with the shortest possible delay (2 seconds). Acquisition parameters were 140 kv and 80 mas eff on the first tube detector system (A system) and 80 kv and 234 mas eff on the second system (B system). Online tube current modulation (CareDose4D, Siemens Healthcare) was enabled. Each detector was collimated to mm with a flying focal spot, and a pitch of 0.65 was applied. Rotation time was set to 0.33 second. Source images (80 and 140 kv) as well as weighted averaged images were reconstructed with a non-edge-enhancing soft reconstruction algorithm (D30); slice thickness and increment were 1.0 and 0.7 mm, respectively. Weighted-averaged images are created by fusing the source images with a weighting factor of 0.4 and resemble regular 120-kV images in terms of dose and image quality [11 13]. A mean volume CT dose index of 7.5 mgy and a mean dose length product of 283 mgy cm were applied. After reconstruction, the images were transferred to a workstation with dedicated commercially available dual-energy postprocessing software (Syngo Dual Energy, Siemens Healthcare). Automatic bone and plaque removal was performed without further manual adjustments of the algorithm. The postprocessing application detects and removes calcified plaques and osseous structure from the data set by examining the ratios of Hounsfield units at 80 and 140 kv for each pixel, which differ between calcium and iodine as a result of the difference in electron density of calcium and iodine. As seen in Figure 1, the attenuation of iodine with an atomic number (Z) of 53 is more energy dependent than is the attenuation of calcium (Z = 20). The relative difference in attenuation at 80 and 140 kv can be expressed using the dualenergy ratio, which is calculated by dividing the Hounsfield units of a certain material at 80 kv by the Hounsfield units at 140 kv. Although the dual-energy ratio is theoretically material specific, practical application of this technique is hampered by image noise, beam hardening, and partial volume effects. To improve the image quality and suppress artifacts, filters such as Blooming Reduction, which reduces the influence of beam hardening, and Fragment Removal, which automatically removes reminiscent fragments of bone marrow, are applied. Bone marrow is difficult to detect with the dual-energy technique because of the mix of iodine-containing bone marrow and spongious bone. In regions with increased image noise, such as the shoulders, threshold-based vessel segmentation techniques are used instead of the dual-energy information. As an output of the postprocessing software, images are created in which all pixels that are identified as bone or calcium are assigned a Hounsfield value of 1,024 HU and thus appear black. These resulting plaque and bone removal images can be used to create thick slab maximum intensity projection images or volumerendered images and can be stored. The postprocessing time was recorded for each case. DSA DSA was performed according to standard procedure guidelines of our institution. A biplanar DSA system (Axiom Artis Zee, Siemens Healthcare) was used. Stenoses were imaged in at least two standardized planes and were quantified according to North American Symptomatic Carotid Endarterectomy Trial criteria [14] by an experienced neuroradiologist. Evaluation The weighted mixed images (without plaque removal) were used to create standardized parasagittal thin-slap maximum intensity projections (3D, Syngo, Siemens; projected slice thickness, 5 mm; increment, 2 mm) and paraaxial multiplanar reformations (slice thickness and increment, 1.5 mm), according to the standard procedure in our institution, referred to hereafter as standard reconstructions. For these standard reconstructions, two experienced neuroradiologists who were blinded to the results of the other evaluations quantified the stenoses in consensus. Three-dimensional evaluation software (In- Space, Siemens) was used to create maximum intensity projections from the plaque and bone removal images (plaque and bone removal maximum intensity projections). Two experienced radiologists who were blinded to the results of DSA and to the standard reconstructions subjectively assessed the grade of the stenoses in consensus, according to North American Symptomatic Carotid Endarterectomy Trial criteria [14]. Although the plaque and bone removal maximum intensity projections can be rotated and scaled freely by the user, only the standard planes that were used for DSA were applied for stenosis quantification. Only the impression of the actual lumen within Fig. 1 Illustration of energy dependence of attenuation of iodine and calcium. Attenuation of iodine and calcium was measured in two images at same slice position, acquired with 140 kv (top) and with 80 kv (bottom). Dual-energy (DE) ratios of calcium and iodine were calculated as explained in Subjects and Methods. AJR:194, June

3 Thomas et al. the stenosis, not additional information such as the extent of vessel contrast beyond the stenosis, was taken into account. Pearson s correlation coefficients were calculated for the results of plaque and bone removal maximum intensity projection stenosis quantification and for stenosis quantification in standard reconstructions versus DSA. Linear regression of the results of plaque and bone removal maximum intensity projection and standard reconstructions versus the results in DSA was performed. Vessel enhancement was measured in the mixed images in axial reformations (slice thickness, 1.5 mm) proximal to the stenoses using a region-of-interest (ROI) method. The sizes of the ROIs were adapted to the diameter of the vessel. For each stenosis, the disagreement between plaque and bone removal maximum intensity projection quantification and DSA was calculated. The influence of vessel enhancement on the performance of plaque and bone removal was assessed by calculating Pearson s correlation coefficient of the disagreement between plaque and bone removal and DSA measurements and the vessel enhancement. A C Fig. 2 Images of lumina in calcified internal carotid artery plaques. Images show digital subtraction angiography (DSA) (left), maximum intensity projections created after plaque and bone removal (middle), and standard maximum-intensity-projection reconstructions (right). A, 73-year-old woman with 50% stenosis correctly identified by plaque and bone removal maximum intensity projection and DSA. B, 82-year-old man with filiform high-grade stenosis, correctly depicted by plaque and bone removal maximum intensity projection and DSA. C, 67-year-old man with filiform stenosis. Residual lumen has been slightly eroded in plaque and bone removal maximum intensity projection but is still visible. D, 64-year-old man with filiform calcified stenosis. Image shows full deletion of residual lumen by plaque and bone removal. Results All CT and DSA examinations were performed successfully. No adverse reactions to contrast medium or other complications occurred. Dual-energy bone removal worked automatically in all patients and did not require any user interaction. The average postprocessing time was 45 seconds, excluding the time for image transfers and saving of the images. Figure 2 shows examples of DSA and plaque and bone removal maximum intensity projection images. DSA revealed 25 stenoses of at least 70% involving the carotid bulb or the internal carotid artery. One of these stenoses was a complete occlusion. Three patients had bilateral stenoses. Three tandem stenoses were found in two patients. Four stenoses were rated lower than 70%. Table 1 summarizes the results of stenosis quantification with plaque and bone removal maximum intensity projection and standard reconstructions in comparison with DSA. After plaque and bone removal, all 25 stenoses of at least 70% including the occlusion and all four stenoses of less than 70% were correctly identified. However, six of the 25 significant stenoses were falsely depicted as occlusions after plaque and bone removal as a result of a deletion of residual lumina in heavily calcified stenoses; these stenoses were rated as 83%, 88%, 94%, 95%, 95%, and 99% in DSA. With standard reconstructions (without plaque and bone removal), no false-positive occlusions were detected; thus, all potentially false-positive findings of occlusions could be rectified using standard reconstructions. Two stenoses that were rated less than 70% in DSA and plaque and bone removal maximum intensity projection evaluation were overestimated as larger than 70% with the standard reconstructions. Figure 3 shows the scatter plots of the correlation of the results of stenoses grading with standard reconstructions and plaque and bone removal maximum intensity projections B D 1592 AJR:194, June 2010

4 TABLE 1: Detection of relevant stenoses and occlusions With plaque and bone removal (PBR) maximum intensity projections (MIPs) and standard reconstructions without PBR in comparison with Digital Subtraction Angiography Disagreement PBR MIPs vs DSA ( %) Result R 2 = PBR MIPs Attenuation (HU) Fig. 4 Plot shows disagreement of stenosis quantification between plaque and bone removal maximum intensity projections (PBR MIPs) and digital subtraction angiography (DSA) versus vessel attenuation. Stenosis > 70% Standard Reconstruction PBR MIPs Occlusion Standard Reconstruction True-positive (no.) False-positive (no.) True-negative (no.) False-negative (no.) Sensitivity (%) Specificity (%) Positive predictive value (%) Negative predictive value (%) Accuracy (%) Degree of Stenosis in PBR MIPs (%) R 2 = Degree of Stenosis in DSA (%) A Degree of Stenosis in Standard Reconstructions (%) R 2 = Degree of Stenosis in DSA (%) Fig. 3 A and B, Plots show correlation of stenosis quantification with plaque and bone removal maximum intensity projections and digital subtraction angiography (DSA) and with standard maximum intensity projection reconstructions and DSA. versus DSA. A Pearson s correlation coefficient of r 2 = was calculated for the correlation of plaque and bone removal MPR stenosis quantification with DSA, whereas stenosis quantification with the standard reconstructions only yielded a correlation coefficient of r 2 = versus DSA. Both correlations were statistically significant (p < and p = , respectively). Linear regression revealed a slope of and an intercept of for plaque and bone removal maximum intensity projection versus DSA and a slope of and an intercept of for standard reconstructions versus DSA. A mean arterial enhancement of 410 HU (range, HU) was reached proximal to the stenoses. A weak, but significant, correlation of the degrees of stenosis in plaque and bone removal maximum intensity projec- B tions and standard reconstructions was found (Pearson s r 2 = ; p = ; Figure 4). Discussion CT angiography is an accepted noninvasive tool for the evaluation of carotid artery disease [3]. Despite its versatile nature, it is limited in the differentiation of calcified structures and contrast media [4]. To overcome this limitation, conventional bone subtraction techniques have been developed that allow the differentiation of bones and vessels on the basis of a second, typically unenhanced, acquisition that is subtracted from the contrast-enhanced scan [15, 16]. In addition, late venous scans can be used as a mask for bone subtraction [17]. Newer subtraction algorithms can automatically register the images and allow a correction of interscan motion artifacts within limits [18]. However, all conventional subtraction techniques are principally limited by the requirement of a second acquisition, which adds additional radiation dose to the examination and may be impaired by motion artifacts. By exploiting the differences in attenuation of iodine and calcium at different x-ray energies due to their different atomic numbers, dual-energy CT has the theoretic potential to distinguish between vessels filled with contrast medium and calcified plaques. Dual-source CT permits the simultaneous acquisition of low- and high-energy data in one examination and thus allows simultaneous imaging without interscan motion and with the application of only a little additional radiation dose [5]. Although the principles of dual-energy CT have been explored for decades [19, 20], the recently introduced dualsource CT scanner is the first CT machine to allow the truly simultaneous acquisition of dual-energy data [5, 21, 22] and thus enables time-dependent applications, such as dualenergy arterial phase scanning [23, 24]. The results of our study show that fully automatic plaque removal in dual-energy carotid artery CT angiography is feasible with only short postprocessing times. Using maximum intensity projection visualization techniques, plaque and bone removal CT luminograms could be produced rapidly. With DSA as a reference standard, plaque and bone removal maximum intensity projections showed a high grade of correlation for the quantification of stenoses and a higher degree of accuracy for the detection of relevant stenoses than standard reconstructions. Also sensitivity, specificity, positive and negative predictive AJR:194, June

5 Thomas et al. values, and accuracy of plaque and bone removal maximum intensity projection readings were 100% each for the detection of relevant stenoses, whereas standard reconstructions were not as reliable (100%, 50%, 93%, 100%, and 93%, respectively). On the other hand, deletions of the residual lumina occurred in six cases (21%) of high-grade stenoses, leading to a severely reduced positive predictive value of 14% and an accuracy of 79% for the detection of occlusions, compared with 100% with standard reconstructions where no falsepositive occlusions were found. Although we categorized these cases as occlusions in our study, it has to be kept in mind that experienced readers would probably not misread these cases because a gap in an artery that otherwise has a normal caliber and is perfused distally is not suggestive of a total occlusion, because back reflux into an occluded artery without a reduction in caliber is extremely unlikely. Nevertheless, the differentiation of high-grade stenoses and occlusions is clinically important, because, in contrast to a high-grade filiform stenosis, an occlusion is not eligible for carotid artery stenting. If treatable stenoses are categorized as occlusions, treatment would be denied to the patients. Thus, it can be concluded that dual-energy plaque and bone removal maximum intensity projections are valuable for the correct evaluation of carotid artery stenoses, but that all pathologic findings in plaque and bone removal images should be verified with standard reconstructions to avoid false-positive results. If confirmation is performed, plaque and bone removal maximum intensity projections can improve the accuracy of CT angiography, because it yielded a higher correlation with DSA than standard reconstructions alone. Additionally, plaque and bone removal maximum intensity projections may be used to gain a rapid overview over the examined volume shortly after the examination. Furthermore, plaque and bone removal maximum intensity projection images might be valuable when presenting vascular findings to referring clinicians or to patients. A recently published similar study with a smaller number of patients (18 analyzed stenoses) has shown significantly different results [10]: The group yielded a much stronger correlation (r 2 = 0.95) between DSA and CT angiography and reported only one false-positive result. However, in that study, only four stenoses were larger than 75% in DSA, which leads to the assumption that the study collective was less morbid than in our study. Taking the results of this study into account, we conclude that dual-energy plaque and bone removal yields better results in less severely stenosed vessels, which are often also less calcified. In these stenoses, the calcified parts are often situated in the periphery of the plaque, whereas the lumen is separated from the calcium by fibrous parts. Thus, the removal of calcified plaques appears less challenging, and the resulting luminogram might probably provide a better ground for reading than standardized reconstructions, where calcifications might alter the impression of the lumen. In the literature, many observations about radiation dose in dual-energy CT can be found. The mean dose length product in our study was 283 mgy cm, which is well within the range of a single energy scanning protocol. For technical reasons, dual-energy CT as performed in the present study goes along with a slightly increased radiation dose compared with a single energy scan with a comparable image quality [11]. However, if it is compared with conventional bone subtraction angiography where two acquisitions are necessary, dual-energy bone removal has the potential to reduce radiation dose, because the second acquisition can be spared [6, 7]. There are several technical factors contributing to the deletions of residual filiform vessel lumina in the case of severe stenoses: First, blooming artifacts caused by beam hardening of the calcified plaques [25] can cover the residual vessel lumen and thus make a differentiation of iodine impossible, especially if the calcified plaque is located in close proximity of the lumen, or they can alter the Hounsfield values of the vessel so that the dual-energy ratio, which is used for the differentiation of iodine and calcium, is altered. Thus, iodine cannot be detected sufficiently. Second, in very small lumina, the image pixels might actually be larger than the vessel, leading to partial volume effects and also aggravating the application of the dual-energy ratio technique. In these cases of small lumina, image noise also plays a larger role. Third, in severely stenosed vessels, the arterial enhancement might be severely reduced, leading to a limited visualization of the affected vessel segments, which is not only an issue for dual energy plaque removal but also for maximum intensity projection visualization in general. It can be expected that the dual-energy segmentation algorithms will be improved in the future and therefore will yield more reliable results, especially if they are combined with vessel tracing and segmentation software. We found a weak correlation between vessel enhancement and the performance of plaque and bone removal in our study. This correlation exists because a high iodine concentration facilitates the differentiation of iodine and calcium. Furthermore, high-contrast images generally produce more meaningful maximum intensity projection images. We recommend the use of high iodine flow settings and the application of optimal bolus timing techniques for dual-energy CT angiography and plaque and bone removal to achieve the best results. In conclusion, dual-energy CT plaque removal can rapidly and fully automatically deliver carotid artery CT angiography luminograms of vessels with calcified plaques with a high correlation to DSA and a high sensitivity for the detection of stenoses and thus has the potential to improve the diagnostic accuracy of CT angiography. However, because of frequent deletions of the residual lumina in calcified plaques, high-grade stenoses might be overestimated as occlusions. Thus, dual-energy plaque and bone removal should be used as a complement to standard reconstructions, not exclusively. A high arterial contrast enhancement is advantageous for a high accuracy. References 1. Josephson SA, Bryant SO, Mak HK, Johnston SC, Dillon WP, Smith WS. Evaluation of carotid stenosis using CT angiography in the initial evaluation of stroke and TIA. Neurology 2004; 63: Chappell FM, Wardlaw JM, Young GR, et al. Carotid artery stenosis: accuracy of noninvasive tests individual patient data meta-analysis. Radiology 2009; 251: Koelemay MJ, Nederkoorn PJ, Reitsma JB, Majoie CB. Systematic review of computed tomographic angiography for assessment of carotid artery disease. Stroke 2004; 35: Saba L, Sanfilippo R, Pirisi R, Pascalis L, Montisci R, Mallarini G. Multidetector-row CT angiography in the study of atherosclerotic carotid arteries. Neuroradiology 2007; 49: Johnson TR, Krauss B, Sedlmair M, et al. Material differentiation by dual energy CT: initial experience. Eur Radiol 2007; 17: Morhard D, Fink C, Graser A, Reiser MF, Becker C, Johnson TR. Cervical and cranial computed tomographic angiography with automated bone removal: dual energy computed tomography versus standard computed tomography. Invest Radiol 2009; 44: Deng K, Liu C, Ma R, et al. Clinical evaluation of 1594 AJR:194, June 2010

6 dual-energy bone removal in CT angiography of the head and neck: comparison with conventional bone-subtraction CT angiography. Clin Radiol 2009; 64: Thomas C, Korn A, Krauss B, et al. Automatic bone and plaque removal using dual energy CT for head and neck angiography: feasibility and initial performance evaluation. Eur J Radiol 2009; [epub ahead of print] 9. Uotani K, Watanabe Y, Higashi M, et al. Dualenergy CT head bone and hard plaque removal for quantification of calcified carotid stenosis: utility and comparison with digital subtraction angiography. Eur Radiol 2009; 19: Watanabe Y, Uotani K, Nakazawa T, et al. Dualenergy direct bone removal CT angiography for evaluation of intracranial aneurysm or stenosis: comparison with conventional digital subtraction angiography. Eur Radiol 2009; 19: Flohr TG, McCollough CH, Bruder H, et al. First performance evaluation of a dual-source CT (DSCT) system. Eur Radiol 2006; 16: Lohan DG, Motamedi K, Chow K, et al. Does dual-energy CT of lower-extremity tendons incur penalties in patient radiation exposure or reduced multiplanar reconstruction image quality? AJR 2008; 191: FOR YOUR INFORMATION 13. Yu L, Primak AN, Liu X, McCollough CH. Image quality optimization and evaluation of linearly mixed images in dual-source, dual-energy CT. Med Phys 2009; 36: North American Symptomatic Carotid Endarterectomy Trial Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Engl J Med 1991; 325: Jayakrishnan VK, White PM, Aitken D, Crane P, McMahon AD, Teasdale EM. Subtraction helical CT angiography of intra- and extracranial vessels: technical considerations and preliminary experience. AJNR 2003; 24: Lell M, Anders K, Klotz E, Ditt H, Bautz W, Tomandl BF. Clinical evaluation of bone-subtraction CT angiography (BSCTA) in head and neck imaging. Eur Radiol 2006; 16: Lell MM, Ditt H, Panknin C, et al. Cervical CT angiography comparing routine noncontrast and a late venous scan as masks for automated bone subtraction: feasibility study and examination of the influence of patient motion on image quality. Invest Radiol 2008; 43: Lell MM, Ditt H, Panknin C, et al. Bone-subtraction CT angiography: evaluation of two different fully automated image-registration procedures for interscan motion compensation. AJNR 2007; 28: Kelcz F, Joseph PM, Hilal SK. Noise considerations in dual energy CT scanning. Med Phys 1979; 6: Michael GJ. Tissue analysis using dual energy CT. Australas Phys Eng Sci Med 1992; 15: Graser A, Johnson TR, Bader M, et al. Dual energy CT characterization of urinary calculi: initial in vitro and clinical experience. Invest Radiol 2008; 43: Meyer BC, Werncke T, Hopfenmuller W, Raatschen HJ, Wolf KJ, Albrecht T. Dual energy CT of peripheral arteries: effect of automatic bone and plaque removal on image quality and grading of stenoses. Eur J Radiol 2008; 68: Thieme SF, Johnson TR, Lee C, et al. Dual-energy CT for the assessment of contrast material distribution in the pulmonary parenchyma. AJR 2009; 193: Yeh BM, Shepherd JA, Wang ZJ, Teh HS, Hartman RP, Prevrhal S. Dual-energy and low-kvp CT in the abdomen. AJR 2009; 193: Choi HS, Choi BW, Choe KO, et al. Pitfalls, artifacts, and remedies in multi-detector row CT coronary angiography. RadioGraphics 2004; 24: Unique customized medical search engine service from ARRS! ARRS GoldMiner is a keyword- and concept-driven search engine that provides instant access to radiologic images published in peer-reviewed journals. For more information, visit AJR:194, June