| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vascular and Interventional Radiology |
1 Departments of Neuroradiology (X.L., J.F.B., T.S.M., J.P.P.)
2 Neurology (O.G., C.L., D.L.), Hôpital Roger Salengro, Boulevard du Professeur Leclercq, 59037 Lille, France.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: In 22 patients, 44 carotid arteries were evaluated with helical CT and selective conventional angiography. CT data were displayed on volume-rendered and maximum intensity projection (MIP) images. Stenoses were measured separately on axial, volume-rendered, and MIP images and on conventional angiograms. Each artery was then graded as having no stenosis, mild (<30%) stenosis, moderate (30%–70%) stenosis, severe (>70%) stenosis, near occlusion, or occlusion.
RESULTS: One case of stenosis was not assessable at axial CT because of an inappropriate scanning plane; four cases were not assessable at MIP CT because of mural calcifications. All carotid arteries were assessable on volume-rendered images despite no depiction of the residual lumen at the site of narrowing in three cases of near occlusion. Correlations between angiography and helical CT were good. Axial, volume-rendered, and MIP images enabled correct classification of stenosis in 88%, 89%, and 90% of arteries, respectively. CT with volume rendering was slightly more sensitive for determining candidates for endarterectomy (ie, those with >70% stenosis and near occlusion); sensitivity was 100% and specificity, 92%.
CONCLUSION: CT angiography with volume rendering enabled accurate evaluation of carotid disease, even when dense calcifications were present. However, no definite advantage over currently available techniques for CT measurement of stenosis severity was found.
Index terms: Carotid arteries, angiography, 1722.1247, 1722.127 • Carotid arteries, CT, 1722.12112, 1722.12115, 1722.12116, 1722.12117 • Carotid arteries, stenosis or obstruction, 1722.7211, 1722.7212, 1722.7213, 1722.7214, 1722.7215 • Computed tomography (CT), maximum intensity projection, 1722.12115 • Computed tomography (CT), volume rendering, 1722.12112, 1722.12115, 1722.12116, 1722.12117
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Ultrasonography (US), the most accessible technique, is a valuable screening tool that provides morphologic and hemodynamic data (5). US examination is sometimes the sole diagnostic procedure performed before endarterectomy; however, it is operator dependent, and additional noninvasive techniques such as magnetic resonance (MR) angiography or computed tomographic (CT) angiography are needed in conjunction with it to select clinically relevant stenoses and to obtain angiogram-like images. MR angiography is a reliable tool for the detection of severe stenoses. However, its tendency to result in an overestimation of the degree of stenosis results in a low positive predictive value (6,7). More recently, a fast gadolinium-enhanced three-dimensional (3D) MR angiographic technique has been reported to enable imaging of a large volume from the arch to the circle of Willis. It allows evaluation of the anterior and posterior circulations (8), but its low spatial resolution prevents the evaluation of carotid bifurcations.
Helical CT with 3D reconstruction produces angiographic images and is accurate in evaluating the degree of ICA stenosis (9–13) despite the radiation dose and need for iodinated contrast material administration. This technique is based on a rapid acquisition of the entire volume owing to the continuous rotation of the gantry and simultaneous displacement of the examination table (14). Data acquisition by using narrow collimation can be reconstructed with overlapping sections; this provides high spatial resolution (15).
Maximum intensity projection (MIP) and shaded surface display are the two main 3D helical CT techniques used to provide information about the complex anatomy of vascular structures (16–18). These two methods have some limitations in the evaluation of ICA stenosis. The main drawback concerns the presence of mural calcifications, which can hinder the visualization of the residual lumen at the level of the stenosis (10,11,13,19). The MIP technique enables the retention of attenuation information but may be hindered by overlapping vessels (11,19). The shaded surface display technique allows visualization of the outer vessel wall and its relationship to the surrounding anatomy, but it offers no information about the residual vessel lumen (10,11,17).
Volume rendering, a third 3D technique (20,21), requires a powerful workstation. This method has some advantages compared with the MIP and shaded surface display techniques. First, the entire CT data set can be incorporated within the 3D image, whereas a small fraction of data is displayed on MIP and shaded surface display images. Second, the number, attenuation, and opacity of voxels can be adjusted separately to allow a change in the transparency of selected materials. Finally, the 3D appearance is maintained and thereby enables a good analysis of the relationship between vascular structures. Clinical applications remained limited for a long time because of computer constraints. Recent advances in graphic software have enabled the Johns Hopkins Medical Institutions, Baltimore, Md, to use this technique in splanchnic and thoracic vascular imaging (22–25). The application of this technique for evaluation of carotid bifurcations may be useful, because the vessel lumen can be evaluated in its transparency in relation to the surrounding structures, irrespective of the presence of arterial wall calcifications. Moreover, owing to the larger data set incorporated in the final image, the degree of carotid stenosis may be defined theoretically with more accuracy compared with that defined with other 3D reconstruction techniques. However, to our knowledge, CT with volume rendering has not yet been evaluated for assessment of carotid arteries.
The purpose of this study was to evaluate helical CT with volume rendering for the assessment of atherosclerotic disease of the extracranial ICA and to compare three CT measurement techniques (ie, axial, volume rendering, and MIP) with selective conventional angiography for estimating the degree of ICA stenosis.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Conventional angiography and helical CT were performed within 3 weeks of each other. Angiograms were obtained on a digital subtraction system (Integris V 3000; Philips Medical Systems, Best, the Netherlands) by using a femoral approach. After arch aortograms with two oblique cervical views were obtained, selective catheterization of the common carotid artery was performed to evaluate the degree of ICA stenosis. The carotid artery studies included at least three projections (posteroanterior, lateral, and ipsilateral 45° oblique), and two images per second were acquired with a 512 x 512 matrix and a 20-cm field of view. For each projection, an 8-mL bolus of iohexol (Omnipaque 300; Nycomed, Princeton, NJ) was injected at a rate of 4 mL/sec by using a power injector. The contrast material dose for the entire study did not exceed 150 mL. The degree of stenosis was determined separately by two observers (X.L., J.F.B.) and was defined as the ratio of the perpendicular diameter of the stenosis at its narrowest point to the normal ICA diameter well cranial to the stenosis. The measurements were made by using a large zoom and a computer caliper with a submillimeter scale. Whenever discordant results were obtained, the involved vessel was reassessed by both observers together to reach a consensus.
Data Acquisition
Helical CT was performed on a Somatom Plus 4 A scanner (Siemens Medical Systems, Erlangen, Germany). Patients were placed in the supine position with the head tilted back to prevent dental artifacts on the images. To determine the level of carotid bifurcations and to detect the presence of calcifications, we first performed sequential scanning in 5-mm sections without contrast material from the C2 through C6 spinal level after acquiring a lateral scout image. For CT angiography, continuous data were acquired over 44 seconds; the scanning commenced at approximately 3 cm proximal to the common carotid bifurcation and proceeded in a cranial manner. A total volume of 120 mL of nonionic contrast material (Omnipaque 350; Nycomed) was administrated at a rate of 3 mL/sec by using a power injector and a 20-gauge intravenous catheter inserted into the antecubital vein. Helical scanning (2-mm collimation, 2-mm/sec table speed, 15-cm field of view, 120 kV, 220 mA, 512 x 512 matrix) was started 20 seconds after the start of the injection. No test bolus of contrast material was used to time the acquisition. Patients were instructed to breathe quietly without swallowing during scanning. The axial images were reconstructed at 1-mm increments by using a 360° linear interpolation algorithm. A high-spatial-resolution convolution algorithm was used in cases of voluminous calcified plaque to differentiate mural calcifications from contrast material. The window level was preset at 150–200 HU, with a width of 400–600 HU.
Volume Rendering
CT data were transferred to an independent workstation (Magic View; Siemens Medical Systems). Regions of interest were selected by manually drawing them on axial source images to exclude bone structures, the enhancing jugular veins, and the distal branches of the external carotid artery. CT angiograms were then reconstructed independently for each side of the neck.
Optimal parameters for generating volume-rendered images of carotid bifurcations were defined by consensus between two senior radiologists (X.L., J.F.B.). The parameters were determined in a preclinical study in which different models of CT with volume rendering were tested. The time required to display 3D images was long owing to the large number of calculations. Each modification of parameter and any change in volume orientation took 2 or 3 minutes. After the initial training period, the protocol was standardized; this enabled technologists to display volume-rendered images after 10–15 minutes of postprocesssing per carotid artery.
Volume-rendered images were generated by selecting trapezoids relative to the voxel intensity histogram. As described by Johnson et al (23), the size and position of the trapezoids determine the number and attenuation of voxels incorporated into the image, and the trapezoid shape influences the gray scale and contrast of the image. These parameters are defined by the window width and window level. The window width is a measurement of the length of the horizontal axis under the upslope, and the window level corresponds to the midpoint of the upslope. We selected two trapezoids to display the contrast material–enhanced arterial lumen (lumen trapezoid) and the mural calcifications (calcification trapezoids). The window width and level were chosen in an empirical manner; the selection of voxels ranged from 250 to 350 HU for the lumen trapezoid and from 670 to 1,030 HU for the calcification trapezoid (Fig 1).
|
Image Analyses
Qualitative assessment was performed in a consensus manner for the following criteria: image contrast, artifact, mural calcification, and ulcer. The parameters of eccentric and concentric stenosis were not evaluated. The image contrast was judged on axial images obtained in each patient at the level of the carotid bifurcations by comparing the attenuation in the enhancing arterial lumen and with that in the surrounding tissue. The contrast was graded as low when the attenuation of the arterial lumen was slightly higher than that of the surrounding muscles, good when the attenuation appeared to be similar to that of the bone structures, and mild in intermediate cases. The analysis of artifacts was that of artifacts related to motion or dental hardware. The mural calcifications on each set of CT images were graded as follows: 0, no calcification; 1, minor calcification; 2, large calcification without consequence to the interpretation of images; or 3, large calcification preventing complete analysis of the residual lumen. An ulcer was defined on the sagittal view as a crater that penetrated the plaque more than 2 mm.
When the artery was judged to be assessable, the arterial lumen was evaluated in a blinded manner by two trained radiologists (X.L., J.F.B.). Axial, MIP, and volume-rendered images were used independently to assess the degree of ICA stenosis; the findings on these three images obtained in each patient were compared between observers, and each was separately compared with the stenosis measured by using angiography. The degree of stenosis was measured by comparing the diameter of the residual lumen at its narrowest point with the diameter of the distal ICA at the level considered to be free of disease and without calcification and to have a normal wall thickness. No measurement was performed when the ICA appeared to be normal, nearly occluded, or occluded. A nearly occlusive situation was defined as a tight stenosis with decreased diameter of the distal ICA. Each carotid artery was then assigned to one of six grades: no stenosis, mild (<30%) stenosis, moderate (30%–70%) stenosis, severe (>70%) stenosis, near occlusion, or occlusion. Cases that led to a disagreement between observers were reviewed by both readers to reach a consensus.
Statistical Analyses
In the first step of analysis, the interobserver agreement on measurement of the percentage of stenosis on each separate set of CT images (ie, axial, MIP, and volume rendered) and on conventional angiograms was evaluated by using the limits of agreement method (26). This test provides an index of the difference between each observer for 95% of observations, or 95% limits of agreement.
In the second step of analysis, the correlation between the stenosis measurement obtained by using the three separate CT imaging techniques (ie, axial, MIP, and volume rendered), with conventional angiography as the reference method, was assessed by using the Pearson test (27,28). The 
statistic was used for stenosis severity categorization (ie, no stenosis, mild stenosis, moderate stenosis, severe stenosis, near occlusion, or occlusion). The values between 0.4 and 0.8 suggested moderate to substantial agreement, and values higher than 0.8, excellent agreement. P values lower than .05 were regarded as significant.
Finally, we assessed the value of each set of CT images to select candidates for endarterectomy. The categorization of obstruction into severe stenosis and near occlusion achieved by using the three CT techniques was compared with that achieved by using conventional angiography among the assessable stenosed arteries with the Fisher exact test (28).
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Qualitative Assessment of Helical CT
All ICAs were correctly depicted within the scanning volume. Contrast enhancement was judged to be mild at the site of carotid bifurcations in four patients and good in the other 18 patients. Motion artifacts related to swallowing or dental hardware were observed in five patients but did not interfere with the stenosis analysis.
Nonenhanced axial images showed minor (grade 1) calcifications in the arterial wall in 24 (54%) of 44 carotid arteries. In seven of 19 arteries with moderate stenosis and in two of three arteries with severe stenosis, calcifications were judged to be large and close to the residual lumen, but this did not interfere with the evaluation of the residual lumen on axial images after contrast material administration (ie, grade 2 calcification). The evaluation was possible in this case because of the use of a large window width and a high-spatial-resolution algorithm, which enabled correct differentiation between the enhancing residual lumen and the surrounding calcifications. One case of severe stenosis was not assessable because of the elongated carotid bulb in the transverse plane, which resulted in a scanning plane that was inappropriate for accurate measurement of stenosis (Fig 2).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Comparison of CT Measurement Techniques
Comparisons between the three techniques are shown in Figure 7 (correlation) and Table 1 (categorization). Correlation and categorization provided different information; similar measurements of stenosis between the three CT techniques could lead to a different classification in cases where the severity of stenosis was close to 30% or 70% and, conversely, a rate of substantial difference in measurement higher than 10% sometimes led to the same classification.
|
|
|
|
For categorization of stenosis severity in the assessable arteries (44 with angiography and CT with volume rendering, 43 with axial CT, and 40 with MIP CT), there was good and significant agreement between the three CT techniques and angiography: On axial images, = 0.90, P < .0001; on volume-rendered images, = 0.85, P < .0001; and on MIP images, = 0.87, P < .0001. However, volume rendering resulted in a misclassification of five (11%) of 44 carotid arteries, and among the assessable arteries, five (12%) were misclassified by using axial images, and four (10%) were misclassified by using MIP images (Table 1). The four nearly occluded ICAs and two occluded ICAs were correctly detected by using all three CT techniques.
Finally, the sensitivity and specificity of helical CT to enable detection of severe stenosis and near occlusion in the assessable arteries was found to be good for the three CT techniques. However, a slight advantage in sensitivity was observed with CT with volume rendering, which enabled correct classification of stenosis severity in the seven arteries with stenosis greater than 70%; MIP imaging enabled correct classification in six, and axial imaging in four of the six assessable arteries (Table 2).
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The 2-mm collimation and pitch of 1 used in the present study enabled accurate evaluation of carotid stenosis. This acquisition protocol is roughly similar to that used in most studies (10–13,19) in which carotid arterial stenosis was evaluated. Dix et al (29), in a recent report in which the accuracy and precision of CT angiography were evaluated in a model of carotid arterial bifurcation stenosis, found no statistically significant difference between a pitch of 1 and of 1.5 or between a collimation of 1.5 mm and of 3 mm. They recommended choosing the larger collimation and pitch because of the greater coverage. Concerning the window level, the results of our study are in accordance with those of Dix et al (29), who demonstrated that the most accurate measurement of the arterial lumen is obtained when the window level is halfway between the lumen attenuation and vessel wall attenuation.
The volume rending technique enabled the acquisition of angiogram-like images with 3D vascular interrelationships and a reliable estimation of the stenosis. The overall agreement with conventional angiography was in accordance with that in most previous reports (11,13,19) in which helical CT in carotid atherosclerotic disease was evaluated, with possible differentiation between nearly occluded arteries and occluded arteries and a high sensitivity and specificity for the determination of candidates for endarterectomy (ie, those with stenosis >70% and near occlusion). The potential advantage of volume rendering compared with shaded surface display and MIP techniques is the possibility to estimate ICA stenosis irrespective of calcifications in the arterial wall. The principle is based on the classification of all voxels on the image according to the probability that the voxel contains a tissue type (ie, a voxel histogram) (20,21). By using a trapezoid function, a material may be selected in the image. Thus, we positioned two trapezoids to isolate enhancing vascular structures and calcifications. We found that with minimum and maximum CT attenuation values of 250–350 HU, the 275-HU level enabled the detection of enhancing vascular lumens in cases of optimal contrast material injection, whereas calcifications were detected by using higher CT attenuation values (ie, 685 HU). We selected a triangular shape for the lumen trapezoid because we found that such a configuration led to a good delineation of the artery outlines, with a homogeneous appearance of the arterial lumen.
As suggested by Johnson et al (25), we attempted to increase the image contrast by using a decreased window width. However, although the rate of correct evaluation of the percentage of ICA stenosis was close to that with conventional angiography, the residual lumen was not depicted at the site of the stenosis in three cases of near occlusion; this might have been because of an excessive increment of the trapezoid slope that led to a decrement of the gray scale. After selecting the position and shape of the trapezoids, the most important parameter to define for each trapezoid was the opacity value. This parameter determines the relative transparency of each selected material needed to differentiate two close structures by using 3D display. Thus, we selected a low opacity value for the calcification and a high opacity value for the enhancing arterial lumen, and this enabled good visualization of the stenosis through the calcifications in all cases.
MIP images enabled correct classification of most stenoses, as previously reported (11,19). However, unlike Link at al (12), we did not find any significant difference in agreement between the different grades of stenosis. As previously reported (11,19), dense arterial wall calcifications were frequent in our study, and this prevented the evaluation of the stenosis in about 10% of cases. Methods to minimize this drawback and to differentiate calcifications from contrast material have been described. The usual technique involves conjointly analyzing the axial source images, which have been proved to be reliable, provided that the window width and the convolution algorithm are appropriate (11). However, the measurement of the stenosis may be inaccurate if the vessel is tortuous and the scanning plane is not perpendicular to the arterial lumen. In our study, the scanning plane was appropriate in most cases and thereby enabled correct evaluation of the stenosis on axial images, except in one carotid artery that was judged to be nonassessable because of a short and severe stenosis oriented in the transverse plane.
Another fast and efficient method to analyze both the vascular lumen and the arterial wall consists of performing multiplanar reformations, but this technique can lead to an overestimation of stenosis severity and, in cases of tortuous vessels, make interpretation of images difficult. Moreover, this method necessitates a proficient and trained operator (9,10). Finally, calcifications can be removed from axial sections by using manual segmentation (10,11) or sophisticated software (13), but this procedure is time-consuming and can result in an overestimation of stenosis severity with the removal of neighboring pixels.
The shaded surface display algorithm technique provides effective 3D rendering with accurate information about vascular interrelationships (10,13). However, this technique was not performed in our study because it has been shown that the shaded surface display algorithm is less effective than the MIP algorithm for evaluating the exact degree of ICA stenosis (11). This can be explained by the arbitrary selection of the threshold, which may lead to an overestimation of stenosis severity if the threshold is too high and, conversely, to an underestimation of stenosis severity if the threshold is low (11,17). Unlike the shaded surface display technique, which involves the use of a binary classification, in the volume rendering technique, the entire data set conveyed through a gray scale can be used to enable a more accurate evaluation of the stenosis.
In conclusion, despite the theoretic interest of volume rendering 3D display over the current postprocessing techniques, no definite advantage of this technique for the evaluation of carotid disease was found. By using the axial sections, the degree of narrowing can be correctly estimated if the scanning plane is roughly perpendicular to the vessel. Mural calcifications constitute a drawback for MIP reconstructions, but the combination of axial and MIP imaging enables an accurate estimation of the degree of stenosis severity in most cases. The use of the volume rendering technique may be useful when dense calcifications are located around or close to the residual lumen, because the arterial lumen may be viewed in its transparency through the surrounding structures. However, further studies in which larger groups are used are needed to determine the exact role of CT angiography with volume rendering for routine evaluation of carotid stenosis.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: ICA = internal carotid artery MIP = maximum intensity projection 3D = three-dimensional
Author contributions: Guarantor of integrity of entire study, X.L.; study concepts and design, J.P.P.; definition of intellectual content, X.L.; literature research, J.F.B., T.S.M.; clinical studies, C.L.; data acquisition, T.S.M.; data analysis, X.L., J.F.B.; statistical analysis, O.G.; manuscript preparation and editing, X.L.; manuscript review, O.G., D.L.
Received April 2, 1998;
revision requested June 5, 1998; revision received July 17, 1998;
accepted September 23, 1998.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. Saba and G. Mallarini MDCTA of Carotid Plaque Degree of Stenosis: Evaluation of Interobserver Agreement Am. J. Roentgenol., January 1, 2008; 190(1): W41 - W46. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. K. Chen, P. T. Johnson, K. M. Horton, and E. K. Fishman Unsuspected Mesenteric Arterial Abnormality: Comparison of MDCT Axial Sections to Interactive 3D Rendering Am. J. Roentgenol., October 1, 2007; 189(4): 807 - 813. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Saba, G. Caddeo, R. Sanfilippo, R. Montisci, and G. Mallarini CT and Ultrasound in the Study of Ulcerated Carotid Plaque Compared with Surgical Results: Potentialities and Advantages of Multidetector Row CT Angiography AJNR Am. J. Neuroradiol., June 1, 2007; 28(6): 1061 - 1066. [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]
|
|
|
|
|
|
M. Lell, C. Fellner, U. Baum, T. Hothorn, R. Steiner, W. Lang, W. Bautz, and F.A. Fellner Evaluation of Carotid Artery Stenosis with Multisection CT and MR Imaging: Influence of Imaging Modality and Postprocessing AJNR Am. J. Neuroradiol., January 1, 2007; 28(1): 104 - 110. [Abstract] [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]
|
|
|
|
