| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
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.
3
Department of Health Evaluation Sciences, A210, Penn State University,
Hershey, PA 17033.
Received September 29, 2000;
accepted after revision April 30, 2001.
Address correspondence to K. D. Hopper.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
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, shaded-surface 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.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CT angiography has been performed using five separate display techniques: axial (cross-sectional), multiplanar reconstruction (MPR), shaded-surface display, maximum intensity projection (MIP), and volume rendering. Although 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.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
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).
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. 2A,2B,2C,2D,2E). 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.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
For all five methods except MPR at 4.0-mm stenosis, the percentage of measurement error for vessel diameters of 6-8 mm was less than 2.5%. For axial, MIP, MPR, and shaded-surface 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.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Several groups of researchers have found that shaded-surface display provides an accurate method of performing CT angiography [2,3,4,5,6,7,8,9,10,11]. For evaluation of the common carotid artery, Schwartz et al. [2], Dillon et al. [3], and Cumming and Morrow [4] found CT angiography shaded-surface display agreed with digital subtraction angiography in 92%, 82%, and 93% of patients, respectively. Similarly, 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 [12,13,14,15,16,17] 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 [10,11,12] 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 angiography 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 shaded-surface 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 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, 25,26,27]. 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, 28,29,30,31,32,33]. 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 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 volume-rendering 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, shaded-surface 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.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Ferencik, D. Ropers, S. Abbara, R. C. Cury, U. Hoffmann, K. Nieman, T. J. Brady, F. Moselewski, W. G. Daniel, and S. Achenbach Diagnostic Accuracy of Image Postprocessing Methods for the Detection of Coronary Artery Stenoses by Using Multidetector CT Radiology, June 1, 2007; 243(3): 696 - 702. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Saba, G. Caddeo, R. Sanfilippo, R. Montisci, and G. Mallarini Efficacy and Sensitivity of Axial Scans and Different Reconstruction Methods in the Study of the Ulcerated Carotid Plaque Using Multidetector-Row CT Angiography: Comparison with Surgical Results AJNR Am. J. Neuroradiol., April 1, 2007; 28(4): 716 - 723. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. P. Villablanca, F. J. Rodriguez, T. Stockman, S. Dahliwal, M. Omura, S. Hazany, and J. Sayre MDCT Angiography for Detection and Quantification of Small Intracranial Arteries: Comparison with Conventional Catheter Angiography Am. J. Roentgenol., February 1, 2007; 188(2): 593 - 602. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 American Society of Interventional & Therapeutic N, Society for Cardiovascular Angiography and Interve, Society for Vascular Medicine and Biology, Society of Interventional Radiology, E. R. Bates, J. D. Babb, D. E. Casey Jr, C. U. Cates, G. R. Duckwiler, T. E. Feldman, et al. ACCF/SCAI/SVMB/SIR/ASITN 2007 Clinical Expert Consensus Document on Carotid Stenting: A Report of the American College of Cardiology Foundation Task Force on Clinical Expert Consensus Documents (ACCF/SCAI/SVMB/SIR/ASITN Clinical Expert Consensus Document Committee on Carotid Stenting) J. Am. Coll. Cardiol., January 2, 2007; 49(1): 126 - 170. [Full Text] [PDF]
|
|
|
|
 |
|
 E. K. Fishman, D. R. Ney, D. G. Heath, F. M. Corl, K. M. Horton, and P. T. Johnson Volume rendering versus maximum intensity projection in CT angiography: what works best, when, and why. RadioGraphics, May 1, 2006; 26(3): 905 - 922. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B.B. Ertl-Wagner, R. Bruening, J. Blume, R.-T. Hoffmann, S. Mueller-Schunk, B. Snyder, and M.F. Reiser Relative Value of Sliding-Thin-Slab Multiplanar Reformations and Sliding-Thin-Slab Maximum Intensity Projections as Reformatting Techniques in Multisection CT Angiography of the Cervicocranial Vessels AJNR Am. J. Neuroradiol., January 1, 2006; 27(1): 107 - 113. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. P. Lawler, H. K. Pannu, and E. K. Fishman MDCT Evaluation of the Coronary Arteries, 2004: How We Do It--Data Acquisition, Postprocessing, Display, and Interpretation Am. J. Roentgenol., May 1, 2005; 184(5): 1402 - 1412. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V A Duddalwar Multislice CT angiography: a practical guide to CT angiography in vascular imaging and intervention Br. J. Radiol., December 1, 2004; 77(suppl_1): S27 - S38. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. K. Kim, J. H. Kim, S.-J. Bae, and K.-S. Cho CT Angiography for Evaluation of Living Renal Donors: Comparison of Four Reconstruction Methods Am. J. Roentgenol., August 1, 2004; 183(2): 471 - 477. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. W. Olin, J. A. Kaufman, D. A. Bluemke, R. O. Bonow, M. D. Gerhard, M. R. Jaff, G. D. Rubin, and W. Hall Atherosclerotic Vascular Disease Conference: Writing Group IV: Imaging Circulation, June 1, 2004; 109(21): 2626 - 2633. [Full Text] [PDF]
|
|
|
|
|
|
B. F. Tomandl, N. C. Kostner, M. Schempershofe, W. J. Huk, C. Strauss, L. Anker, and P. Hastreiter CT Angiography of Intracranial Aneurysms: A Focus on Postprocessing RadioGraphics, May 1, 2004; 24(3): 637 - 655. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Mallouhi, M. Rieger, B. Czermak, M. C. Freund, P. Waldenberger, and W. R. Jaschke Volume-Rendered Multidetector CT Angiography: Noninvasive Follow-Up of Patients Treated with Renal Artery Stents Am. J. Roentgenol., January 1, 2003; 180(1): 233 - 239. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Corti, M. Alerci, R. Wyttenbach, P. L. Pedrazzi, A. Gallino, T. Hirai, and Y. Korogi Usefulness of Multiplanar Reconstructions in Evaluation of Carotid CT Angiography * Drs Hirai and Korogi respond: Radiology, January 1, 2003; 226(1): 290 - 292. [Full Text] [PDF]
|
|
|
|
|
|
D. D. Cody AAPM/RSNA Physics Tutorial for Residents: Topics in CT: Image Processing in CT RadioGraphics, September 1, 2002; 22(5): 1255 - 1268. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Hochmuth, U. Spetzger, and M. Schumacher Comparison of Three-Dimensional Rotational Angiography with Digital Subtraction Angiography in the Assessment of Ruptured Cerebral Aneurysms AJNR Am. J. Neuroradiol., August 1, 2002; 23(7): 1199 - 1205. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
