CT Angiography

Downloaded from www.ajronline.org by 104.162.232.1 on 06/12/15 from IP address 104.162.232.1. Copyright ARRS. For personal use only; all rights reserved

CT Angiography: In Vitro Comparison of Five Reconstruction Methods Kimberly A. Addis 1 Kenneth D. Hopper 2 Tunç A. Iyriboz 2 Yi Liu 2 Scott W. Wise 2 Claudia J. Kasales 2 Judy S. Blebea 2 David T. Mauger 3

OBJECTIVE. Five image reconstruction techniques have been used with CT angiography: axial (cross-sectional), maximum intensity projection (MIP), curved multiplanar reconstruction (MPR), shaded-surface display, and volume rendering. This study used a phantom to compare the accuracy of these techniques for measuring stenosis. SUBJECTS AND METHODS. A 19-vessel phantom containing various grades of concentric stenoses (0–100%) and three lengths (5, 7.5, and 10 mm) of stenoses was used for this study. Scans were obtained with a slice thickness of 2.0 mm, slice interval of 1.0 mm, pitch of 1.0, 120 kVp, 200 mA, and with the vessels oriented parallel to the z-axis and opacified with nonionic contrast material. CT angiography images were produced using five optimized techniques: axial, MIP, MPR, shaded-surface display, and volume rendering; and measurements were made with an electronic cursor in the normal lumen and mid stenosis by five separate investigators who were unaware of vessel and stenosis diameters. Each of the techniques was first optimized according to the radiology literature and our own preliminary testing. RESULTS. For vessels greater than 4 mm in diameter, axial, MIP, MPR, shaded-surface display, and volume-rendering CT angiography techniques all had a measurement error of less than 2.5%. However, axial, MIP, MPR, and shaded-surface display techniques were less accurate in estimating smaller (≤ 4 mm) diameters. Volume rendering tended to be more accurate in the measurement of vessels with a 2.0- to 4.0-mm diameter and was statistically more accurate for diameters of 0.5–1.0 mm ( p < 0.001). CONCLUSION. All five CT angiography display techniques (axial, MIP, MPR, shadedsurface display, and volume rendering) accurately display vessels and stenoses greater than 4 mm in diameter. However, volume rendering tends to be more accurate for stenoses of 2–4 mm and was statistically better in the measurement of diameters of 0.5–1.0 mm ( p < 0.001). Volume rendering is an accurate method for evaluating all grades of stenoses.

C

Received September 29, 2000; accepted after revision April 30, 2001. 1 College of Medicine, Penn State University, Hershey, PA 17033. 2 Department of Radiology, H066, Penn State University, P. O. Box 850, Hershey, PA 17033. Address correspondence to K. D. Hopper. 3

Department of Health Evaluation Sciences, A210, Penn State University, Hershey, PA 17033. AJR 2001;177:1171–1176 0361–803X/01/1775–1171 © American Roentgen Ray Society

AJR:177, November 2001

T angiography is becoming an accepted, accurate method to detect cerebral aneurysms and to measure stenoses of the carotid and renal arteries. Using only IV contrast material and incurring fewer complications and requiring less time and less expense than conventional angiography, CT angiography is becoming increasingly popular. This is especially true with the widespread availability of helical CT for patients whose symptoms might not warrant the invasiveness of conventional angiography, and as a followup in after surgery and other procedures. CT angiography has been performed using five separate display techniques: axial (crosssectional), multiplanar reconstruction (MPR), shaded-surface display, maximum intensity projection (MIP), and volume rendering. Al-

though all these techniques are useful in displaying CT angiographic data, considerable disagreement has arisen in the radiology literature as to the accuracy of each method and which technique is optimal. To address this disagreement, our study evaluated these five techniques using a phantom containing stenoses of known diameters. Each of the five techniques was first optimized according to the radiology literature and our own preliminary testing. Subjects and Methods A dedicated vascular phantom (Rando; The Phantom Laboratory, Salem, NY) with an attenuation of 22 H and a diameter of 17 cm was used for this study. The phantom was designed to simulate the carotid vasculature and had a total attenuation comparable to a human neck. The normal luminal diameter of each

1171


vessel was 8 mm. Stenoses of 0%, 25%, 50% 75%, 87.5%, and 93.8% were placed in the center of these vessels. Stenoses of each percentage were constructed in 5-, 7.5-, and 10-mm lengths. All phantom vessels and their stenoses were manufactured with an accuracy of 1 µm, as determined by a micrometer. One 10mm-long occlusion was included as well. The angle of inclination from the normal lumen into the stenosis was a constant 75º. Each stenosis was concentric and centered on the normal lumen. Nonionic contrast material in a 1-L container was carefully diluted with distilled water until an attenuation of 250 H was achieved on repeated CT scanning (attenuation difference of background vs contrast material, 228 H). The phantom vessels were then filled with this diluted contrast material, and all air bubbles were removed. The phantom was scanned (PQ5000; Marconi Medical Systems, Highland Heights, OH) with a 17-cm field of view, 2-mm slice thickness, 1.0-mm reconstruction interval, 175 mA, 150 kVp, 1.0 pitch, and standard 180º interpolator and reconstruction algorithms. The phantom vessels were scanned oriented along the long axis of the scanner parallel to the z-axis. For shaded-surface display, a lower display threshold of 100 H was used, which has been established as the optimal threshold with this contrast-to-background density difference [1]. For axial (perpendicular to vessel lumen) and MIP images, the display window and level were optimized for this particular contrast density. Similarly, for MPR and axial images, measurements from one inside vessel wall and contrast edge to the opposite inside wall and contrast edge were obtained. For MPR, a 0.6-mm reconstructed image plane was created that could be moved dynamically through the volume at 0.3-mm intervals. Volume-rendered images had measurements taken from one inside wall to the opposite inside wall (Fig. 1). To optimize volume rendering, the transition between the contrast lumen and the vessel wall was reconstructed separately (Table 1). The attenuation coefficient of this transition thickness of 2–3 voxels was set as: attenuation coefficient vessel contrast (H) + soft-tissue background (H) / 2. The window for volumetric display was defined as the range between the vessel contrast and the soft-tissue attenuation (250 H – 22 H = 228 H).

TABLE 1

Addis et al.

Optimized Tool Settings for Volume-Rendering Technique

Setting No.

Density Range (H)

Opacity (%)

Corresponding Phantom Region

I II III IV

–1000 to 45 112–113 180–1900 2000–3000

0 75 50 85

Soft tissue Transition zone (endothelium) Contrast-filled vessel lumen High-density lesions (e.g., calcifications)

Five experienced investigators who were unaware of the actual vessel and stenosis diameters worked independently and measured the vessel diameters with each of the five display techniques. Measurements were performed in the same order (axial, MPR, shaded-surface display, MIP, volume rendering). The reviewers were able to electronically manipulate the images for all five display techniques to optimize measurement of the stenosis and normal vessel diameter for each vessel. For the axial measurements, the reviewers scrolled through the axial image to select the optimal location to measure the diameter of both the mid stenosis and the mid normal lumen. For MPR, sequential coronal 0.6-mm slices (reconstruction interval, 0.3 mm) were created by the reviewer, who selected the optimal slice for measurement. MIP images were created in the coronal plane. The shaded-surface display and volume-rendered images were rotated so that measurements could be done with the vessel displayed longitudinally (Fig. 2). A magnification of 8 on a black background was used for all measurement techniques to improve the visualization of the vessel edge. Electronic calipers were used, allowing values to be recorded to within 0.1 mm. Each measurement was done twice and averaged. Descriptive statistics including mean, standard deviation, and number of false occlusions were calculated for each method and stenosis diameter combination. Analysis of variance was used to test separately for differences among the five reconstruction methods for each grade of stenosis. Pairwise comparisons among individual methods were made using the Tukey multiple comparison correction method. Box plots of the absolute errors were used to provide a visual comparison of the reconstruction methods at each grade of stenosis and to assess the relationship between absolute measurement error and vessel diameter for each reconstruction method. The absolute error of each measurement was calculated as the difference between the measured value and the known value. To assess the potential effect of interobserver variation of these results, F tests were used to test for differences among observers at the optimized point for each reconstruction method.

Results

Fig. 1.—For volume rendering, all measurements were made from inside of one vessel wall to inside of opposite wall (arrow ).

1172

The mean diameter, number of measurements, number of false occlusions, and standard deviations are listed in Table 2 and Figures 3–5 for all five CT angiography display techniques. For all five methods except MPR at 4.0mm stenosis, the percentage of measurement

error for vessel diameters of 6–8 mm was less than 2.5%. For axial, MIP, MPR, and shadedsurface display techniques, measurement accuracy decreased significantly (10–25%) at diameters of 4.0 mm and smaller (stenosis of 4.0 mm, p = 0.006; 2.0 mm, p = 0.06; 1.0 mm, p = 0.0006; 0.5 mm, p = 0.0001). However, for stenosis diameters of 4.0 mm and smaller, measurements using volume rendering were within 10% of the known value. Although volume rendering tended to be more accurate for 2- and 4-mm stenoses, it had statistically less measurement error ( p < 0.001) than the other four techniques (Fig. 3) for stenoses of 0.5 and 1.0 mm. For instance, the mean vessel diameters measured in 1.0- and 0.5-mm stenoses by volume rendering were 0.95 and 0.47 mm, respectively, with no false occlusions seen. Similar measurements for the axial technique (perpendicular to the vessel lumen) were 0.44 and 0.00 mm, respectively, with 65% false occlusions. With MIP, the mean measured diameter for 1.0- and 0.5-mm stenoses were 0.49 and 0.00 mm, respectively, with 50% false occlusions. For MPR, the mean measurements were 0.41 and 0.00 mm, respectively, with 50% false occlusions. Lastly, for shaded-surface display, means for 1.0- and 0.5-mm stenoses were 0.5 and 0.15 mm, respectively, with 22% false occlusions. Mean measurement error is represented for all five reconstruction methods and for diameters as a box plot (Figs. 3 and 4). All methods provided excellent approximations for the known values for larger diameters (4.0–8.0 mm). With the exception of MPR at 4.0 mm, all the boxes in this range straddle the zero error line. There is a transition at the 2.0-mm diameter, where the error values for axial, MIP, MPR, and shaded-surface display techniques significantly worsen (axial, p = 0.0001; MIP, p = 0.0001; MPR, p = 0.001; shaded-surface display, p = 0.006). Only volume rendering maintained an accurate estimation of the true diameter for highly stenotic vessels. An equality of variance F test was performed to assess interobserver variability and revealed no statistical difference among observers across all diameters for all five techniques. No statistical difference was noted for the length of stenosis.


Reconstruction Methods for CT Angiography Discussion


Five separate techniques have been used in the performance of CT angiography: axial, MPR, shaded-surface display, MIP, and volume rendering. Although shaded-surface disTABLE 2

each of these reconstruction methods and which method is the best. Several groups of researchers have found that shaded-surface display provides an accurate method of performing CT angiography

play and MIP are the most commonly used reconstruction methods for CT angiography, the other three methods are also frequently used. Unfortunately, disagreement exists in the radiology literature as to the accuracy of

Comparison of Five Optimized Reconstruction Methods

Gold Standard Vessel Diameter

Maximum Intensity Projection

x

Error

SD

FO/ n

x

Error

SD

FO/ n

x

Error

SD

FO/ n

x

Error

SD

FO/ n

x

8 mm 6 mm 4 mm 2 mm 1 mm 0.5 mm

8.03 6.06 4.04 1.73 0.44 0.00

a a a c d d

0.16 0.17 0.11 0.20 0.39 —

0/190 0/30 0/30 0/30 9/30 30/30

8.02 6.00 3.90 1.72 0.49 0.00

a a a c d d

0.14 0.08 0.13 0.14 0.17 -—

0/190 0/30 0/30 0/30 0/30 30/30

7.95 5.92 3.81 1.54 0.41 0.00

a a b c d d

0.17 0.17 0.15 0.31 0.23 —

0/190 0/30 0/30 0/30 0/30 30/30

7.98 6.03 3.93 1.76 0.5 0.15

a a a c d d

0.18 0.13 0.15 0.23 0.40 0.27

0/95 0/15 0/15 0/15 2/15 11/15

8.00 6.00 3.99 1.82 0.95 0.47

Axial

Multiplanar Reconstruction

Shaded-Surface Display

Volume Rendering Error SD a a a b b b

FO/ n

0.20 0/247 0.18 0/36 0.24 0/36 0.16 0/36 0.26 0/36 0.25 0/36

Note.—Data are mean diameter measurements (mm) at optimized point (x) and number of false occlusions (FO) / number of data points ( n). Gold standard is manufacturer’s measurement of 2 ± 1 µm. Dash (—) indicates not applicable. a Error from gold standard of < 2.5%. b Error from gold standard of > 2.5% and < 10%. c Error from gold standard of > 10% and < 25%. d Error from gold standard of > 25%.

A

B

C

D

E

Fig. 2.—This example from our study (stenosis diameter, 4.0 mm; diameter of normal vessel lumen on each side of stenosis, 8.0 mm; stenosis length, 10.0 mm) was reconstructed by each of five techniques studied. A–E, CT angiography images of the same vessel using axial (perpendicular to vessel lumen) (A), maximum-intensity-projection (B), multiplanar reconstruction (C), shaded-surface display (D), and volume-rendering (E) techniques. Sample measurements are included on each image.


1173


Addis et al.

Fig. 3.—Box plot displays absolute error (in millimeters) for five optimized reconstruction techniques for each vessel diameter. Each box encloses middle 50% of data points (i.e., from first quartile to third). Median is marked by horizontal line in each box. Lines extend above and below each box to most extreme data point not more than 1.5 times interquartile range from edge of box. Data points farther from box are individually marked. MIP = maximum intensity projection, MPR = multiplanar reconstruction, SSD = shaded-surface display.

Fig. 4.—Box plot displays absolute error (in millimeters) for vessel diameters for each optimized reconstruction technique. AX = axial, MIP = maximum intensity projection, MPR = multiplanar reconstruction, SSD = shadedsurface display, Vol = volume rendering.

[2–11]. For evaluation of the common carotid artery, Schwartz et al. [2], Dillon et al. [3], and Cumming and Morrow [4] found CT an1174

giography shaded-surface display agreed with digital subtraction angiography in 92%, 82%, and 93% of patients, respectively. Simi-

larly, in the assessment of intracranial aneurysms, Liang et al. [5] found shaded-surface display to be 88% sensitive in detecting aneurysms of 2 mm or larger. Dix et al. [6], evaluating CT angiography MIP versus shaded-surface display using a phantom, found a measurement accuracy within 1% using both methods. Other groups of investigators [12–17] have obtained equally good results by reconstructing CT angiography data with MIP. Halpern et al. [12] found that in the renal arteries CT angiography with MIP as compared with digital subtraction angiography had a 96% and 88% sensitivity and specificity, respectively. Marks et al. [13], in evaluating 28 carotid bifurcations with atherosclerotic disease, found an 89% agreement between CT angiography MIP and digital subtraction angiography. In three cases, however, calcifications around the carotid bifurcation interfered with CT angiography MIP evaluation. In evaluating 40 atherosclerotic carotid arteries, Leclerc et al. [17] found CT angiography with MIP correlated well with CT angiography in 96% of cases. Unfortunately, 10 of 40 carotid arteries could not be evaluated with CT angiography MIP because of overlying vascular calcifications or other reasons. To overcome these variable results, some groups of researchers [10, 11, 16] combined shaded-surface display and MIP to achieve diagnostic results. Beregi et al. [16] and Elkohen et al. [10] obtained similar sensitivities and specificities of 88% and 98%, respectively, when evaluating the renal artery, as compared with digital subtraction angiography using such a technique. Similarly, Raptopoulos et al. [11] found the combined use of shaded-surface display and MIP, as compared with digital subtraction angiography, resulted in a sensitivity and specificity of 93% and 96%, respectively, for the detection of extensive aortoiliac disease. Some researchers have found shaded-surface display to be superior to MIP when both are compared with digital subtraction angiography. In the assessment of cerebral aneurysms, Kallmes et al. [8] found CT angiography with shaded-surface display far superior to MIP ( p < 0.001). In the assessment of 22 aneurysms, Davros et al. [9] also found shaded-surface display to be superior to MIP, especially when the rendering threshold of –50 H was used. However, other researchers [10–12] have found opposite results. Galanski et al. [18] found that both shaded-surface display and MIP produced inferior results to axial and MPR CT angiography. Leclerc et al. [17] found CT anAJR:177, November 2001


Reconstruction Methods for CT Angiography giography with shaded-surface display was only 79% accurate versus 95% and 96% for the axial and MIP techniques, respectively. In the renal artery, Rubin et al. [15] found shaded-surface display only 55% accurate in significant grade 2 (70–99%) or 3 (100%) stenoses versus 80% for MIP. Castillo and Wilson [19] and Castillo [20] found that MIP of the carotid bifurcation did not compare well with digital subtraction angiography, having only a 50–67.5% agreement. In the evaluation of 20 patients (40 bifurcations), Castillo and Wilson found that CT angiography with MIP overestimated stenoses in 11 of 20 patients and missed two plaque ulcerations. The use of axial and multiplanar reconstructions, either by themselves or in combination with each other or with other CT angiography reconstruction methods, has been found to correlate well with digital angiography. In the carotid artery, Leclerc et al. [21] found axial images alone had a 95% correlation with digital subtraction angiography in 40 carotid bifurcations. These researchers found the same accuracy as with MIP (96%), although the axial images could evaluate the 10 bifurcations not assessable by MIP. Both techniques were superior to shaded-surface display, for which they found only a 79% correlation to digital subtraction angiography. Brink et al. [22] compared the use of axial, MPR, and MIP CT angiography reconstruction methods with digital subtraction angiography in the renal artery. They found that MIP plus the axial images provided the most consistent results. In 44 carotid bifurcations, Leclerc et al. found axial images alone were as accurate as MIP and were proven superior in five bifurcations where prominent calcified plaque interfered with the MIP reconstructions. In 54 renal arteries in 22 patients, Galanski et al. [18] found that axial combined with MPR images were superior to shadedsurface display and MIP images as compared with digital subtraction angiography. Three groups of investigators [4, 10, 11] combined MPR and shaded-surface display images to best evaluate vascular structures on CT angiography. Compared with digital subtraction angiography, those authors found 93%, 98%, and 85% correlations in the carotid, renal, and aortoiliac vessels. Kaatee et al. [23] combined MPR and MIP CT angiography images in evaluating renal artery stenosis. In 71 patients (166 renal arteries), for stenosis grades 1–3, they found a sensitivity of 92–100% and a specificity of 96–100% as compared with digital subtraction angiography. On the other hand, Berg et AJR:177, November 2001

Fig. 5.—Chart displays comparison of five reconstruction methods at various vessel diameters and shows standard error for each. MPR = multiplanar reconstruction, MIP = maximum intensity projection, SSD = shaded-surface display, Vol Rend = volume rendering.

al. [24] found that the addition of MPR to MIP images in the renal artery improved accuracy, compared with digital subtraction angiography, from 60–75% to 84%. Initial work has been performed in the use of volume rendering with CT angiography [21, 25–27]. Volume rendering provides a composite three-dimensional image by combining MIP images from multiple directions. Although three of these studies are demonstrative (ie., they describe the method but do not present supporting data), Leclerc et al. [21] directly compared axial and MIP images with volumetric CT angiography in the carotid arteries and found equivalent results among the three techniques. The disagreement that has occurred in the results possible with CT angiography with each of these five techniques may have occurred because of an incomplete understanding of the imaging requirements of each method and how to optimize each method. The optimum scanning techniques necessary to achieve good results have also required research, and considerable work has been done

in this regard [1, 28–33]. Building on the past efforts of many investigators to optimize the scanning parameters and display methods for each of the five reconstruction techniques, we executed a study that is, to our knowledge, the first to provide a thorough in vitro comparison of all five methods. We attempted to thoroughly compare axial, MIP, MPR, shaded-surface display, and volume-rendering techniques simultaneously with known values and with each other. Our investigation benefited from much of the research that addressed the issues surrounding optimization of each display technique. Clearly, the first step toward accurate measurement is optimization of the CT angiography display technique being evaluated, and failure to do this may have led to conflicting results by previous studies. Our study shows that using optimized settings, all five CT angiography methods estimate stenosis diameters greater than 4.0 mm with less than 2.5% error. For higher grade stenoses, CT angiography with volume rendering tended 1175


Addis et al. to provide more accurate measurements and resulted in no false-positive occlusions. For stenoses of less than 4.0 mm, the mean percentage of error was less than 10% for volume rendering versus 25% or more for the other four techniques. Volume rendering was statistically better than the other four reconstruction techniques for severe stenoses (diameters of 0.5–1.0 mm). Our technique had several limitations. First, we used static contrast material for imaging. In vivo, blood is flowing and the vessel pulsating, factors that affect measurement accuracy. In addition, other patient motion, such as breathing and swallowing, can affect measurement accuracy. The effect of these variables on measurement accuracy was not evaluated in this study. Although unaware of the true stenosis diameter, each reviewer performed his or her measurements in the same order. This procedure may have injected some bias into the study. However, the use of a proven phantom (Rando) with known vessel and stenosis diameters and the absence of motion artifacts, which could deleteriously affect measurement accuracy, offers the ideal model for comparing the accuracy of multiple reconstruction methods. Postacquisition processing for the volumerendering images was minimal. The technique is quickly learned and easily optimized for each case. Software provided on many CT workstations allows additional image enhancement with the use of color, which may improve image quality even further. In summary, any of the five optimized CT angiography techniques (axial, MIP, MPR, shadedsurface display, or volume rendering) can be used to accurately measure vessel diameters of 4.0 mm and greater. When diameters are less than 4.0 mm, volume rendering was the most accurate, especially for stenoses of 0.5–1.0 mm. Volume rendering is quick, flexible, and the most accurate method to evaluate all grades of stenosis. Although the data from this study are from a phantom, they are directly applicable to clinical imaging and may be used to achieve continued improvements in image quality and to advance the use of CT angiography.

References 1. Liu Y, Hopper KD, Mauger D, Addis KA. CT angiographic measurement of carotid artery: optimizing visualization by manipulating window, level

1176

and contrast density. Radiology 2000;217:494–500 2. Schwartz RB, Jones KM, Chernoff DM, et al. Common carotid artery bifurcation: evaluation with spiral CT. Radiology 1992;185:513–519 3. Dillon EH, van Leeuwen MS, Fernandez MA, Eikelboom BC, Mali WP. CT angiography: application to the evaluation of carotid artery stenosis. Radiology 1993;189:211–219 4. Cumming MJ, Morrow IM. Carotid artery stenosis: a prospective comparison of CT angiography AJR and conventional angiography. 1994;163:517–523 5. Liang EY, Chan M, Hsiang JHK, et al. Detection and assessment of intracranial aneurysms: value of CT angiography with shaded-surface display. AJR 1995;165:1497–1502 6. Dix JE, Evans AJ, Kallmes DF, Sobel AH, Phillips CD. Accuracy and precision of CT angiography in a model of carotid artery bifurcation stenosis. AJNR 1997;18:409–415 7. Rubin GD, Dake MD, Napel SA, McDonnell CH, Jeffrey RB. Three-dimensional spiral CT angiography of the abdomen: initial clinical experience. Radiology 1993;186:147–152 8. Kallmes DF, Evans AJ, Woodcock RJ, et al. Optimization of parameters for the detection of cerebral aneurysms: CT angiography of a model. Radiology 1996;200:403–405 9. Davros WJ, Obuchowski NA, Berman PM, Zeman RK. A phantom study: evaluation of renal artery stenosis using helical CT and 3D reconstructions. J Comput Assist Tomogr 1997;21:156–161 10. Elkohen M, Beregi JP, Deklunder G, Artaud D, Mounier-Vehier C, Carre AG. A prospective study of helical computed tomography angiography versus angiography for the detection of renal artery stenoses in hypertensive patients. J Hypertens 1996;14:525–528 11. Raptopoulos V, Rosen MP, Kent KC, Kuestner LM, Sheiman RG, Pearlman JD. Sequential helical CT angiography of aortoiliac disease. AJR 1996;166:1347–1354 12. Halpern EJ, Rutter CM, Gardiner GA, et al. Comparison of Doppler US and CT angiography for evaluation of renal artery stenosis. Acad Radiol 1998;5:524–532 13. Marks MP, Napel S, Jordan JE, Enzmann DR. Diagnosis of carotid artery disease: preliminary experience with maximum-intensity-projection spiral CT angiography. AJR 1993;160:1267–1271 14. Napel S, Marks MP, Rubin GD, et al. CT angiography with spiral CT and maximum intensity projection. Radiology 1992;185:607–610 15. Rubin GD, Dake MD, Napel S, et al. Spiral CT of renal artery stenosis: comparison of three-dimensional rendering techniques. Radiology 1994;190:181–189 16. Beregi J-P, Elkohen M, Deklunder G, Artaud D, Coullet JM, Wattinne L. Helical CT angiography compared with arteriography in the detection of renal artery stenosis. AJR 1996;167:495–501 17. Leclerc X, Godefroy O, Pruvo JP, Leys D. Computed tomographic angiography for the evaluation of carotid artery stenosis. Stroke 1995;26:1577–1581

18. Galanski M, Prokop M, Chavan A, Schaefer CM, Jandeleit K, Nischelsky JE. Renal arterial stenoses: spiral CT angiography. Radiology 1993;189:185–192 19. Castillo M, Wilson JD. CT angiography of the common carotid artery bifurcation: comparison between two techniques and conventional angiography. Neuroradiology 1994;36:602–604 20. Castillo M. Diagnosis of disease of the common carotid artery bifurcation: CT angiography vs catheter angiography. AJR 1993;161:395–398 21. Leclerc X, Godefroy O, Lucas C, et al. Internal carotid artery stenosis: CT angiography with volume rendering. Radiology 1999;210:673–682 22. Brink JA, Lim JT, Wang G, Heiken JP, Deyoe LA, Vannier MW. Technical optimization of spiral CT for depiction of renal artery stenosis: in vitro analysis. Radiology 1995;194:157–163 23. Kaatee R, Beek FJ, de Lange EE, et al. Renal artery stenosis: detection and quantification with spiral CT angiography versus optimized digital subtraction angiography. Radiology 1997;205:121–127 24. Berg MH, Manninen HI, Vanninen RL, Vainio PA, Soimakallio S. Assessment of renal artery stenosis with CT angiography: usefulness of multiplanar reformation, quantitative stenosis measurements, and densitometric analysis of renal parenchymal enhancement as adjuncts to MIP film reading. J Comput Assist Tomogr 1998;22:533–540 25. Kuszyk BS, Heath DG, Ney DR, et al. CT angiography with volume rendering: imaging findings. AJR 1995;165:445–448 26. Johnson PT, Heath DG, Kuszyk BS, Fishman EK. CT angiography with volume rendering: advantages and applications in splanchnic vascular imaging. Radiology 1996;200:564–568 27. Rubin GD, Paik DS, Johnston PC, Napel S. Measurement of the aorta and its branches with helical CT. Radiology 1998;206:823–829 28. Wise SW, Hopper KD, Schwartz TA, Ten Have TR, Kasales CJ. Technical factors of CT angiography studied with a carotid artery phantom. AJNR 1997;18:401–408 29. Wise SW, Hopper KD, Ten Have T, Schwartz T. Measuring carotid artery stenosis using CT angiography: the dilemma of artifactual lumen eccentricity. AJR 1998;170:919–923 30. Claves JL, Wise SW, Hopper KD, Tully D, Ten Have TR, Weaver J. Evaluation of contrast densities in the diagnosis of carotid stenosis by CT angiography. AJR 1997;169:569–573 31. Diederichs CG, Keating DP, Glatting G, Oestmann JW. Blurring of vessels in spiral CT angiography: effects of collimation width, pitch, viewing plane and windowing in maximum intensity projection. J Comput Assist Tomogr 1996;20:965–974 32. Takahashi M, Ashtari M, Papp Z, et al. CT angiography of carotid bifurcation: artifacts and pitfalls in shaded-surface display. AJR 1997;168:813–817 33. Vandermeulen D, Van Hoe L, Marchal G, Suetens P, Baert AL. Assessment of blood vessel diameter measurements in spiral CT. (abstr) Radiology 1994;193(P):171
