Laboratory Investigation Coronary Artery Disease

The absolute cross-sectional area of a coronary stenosis measured by quantitative coronary angiography correlates well with its hemodynamic significance. We evaluated a combined approach using edge detection applied to the normal segment and videodensitometry applied to the stenosis to determine the absolute cross-sectional area of the stenosis (videodensity method). The results were then compared with those with the edge detection method applied directly to the stenosis. The area of the stenosis by the edge detection method was calculated by analyzing two orthogonal projections for irregular stenoses and with use of the formula for the area of an ellipse (ellipse method). The accuracy of both these techniques was assessed by analyzing digital angiograms acquired from closed-chest dogs in which 10 plastic cylinders with precisely machined circular and irregular lumina were inserted into the coronary arteries. Angiograms of irregular stenoses were acquired in two orthogonal views. The ellipse method applied to circular stenoses was very accurate, with r = .97, average absolute difference (AAD) = 0.21 mm2, and SEE = 0.30. For the videodensity method r = .97, AAD= 0.84 mm2, and SEE = 0.40. Irregular stenoses were better quantitated by the videodensity method applied in one view (AAD = 0.50 mm2, SEE = 0.47) than by the ellipse method applied in two orthogonal projections (AAD = 1.03 mm2, SEE = 0.87). Overall, the two methods were comparable in accuracy (for videodensity, AAD = 0.65 mm2, SEE = 0.71 vs AAD = 0.54 mm2, SEE = 0.79 for ellipse). Although the ellipse method is more accurate for circular stenoses, the videodensity method has the advantage of quantitating the full range of shapes of stenosis in only one projection. Circulation 74, No. 5, 1099-1106, 1986. ANGIOGRAPHIC ASSESSMENT of the percent narrowing of a coronary arterial stenosis has been traditionally accepted as the standard for the evaluation of coronary disease. Yet, studies have shown that the physiologic significance of a coronary arterial stenosis correlates poorly with the percent stenosis.2-5 Recently, Harrison et al.6 showed that the absolute crosssectional area of a coronary stenosis is a better determinant of the physiologic significance of an obstruction.6 A number of different methods have been developed to determine the absolute cross-sectional area of a coronary stenosis seen on the angiogram. Some investigators manually traced the edges of coronary arteries From the Division of Cardiology, Long Island Jewish Medical Center, New Hyde Park, NY. Address for correspondence: Joseph Wiesel, M.D., The Heart Institute, Long Island Jewish Medical Center, New Hyde Park, NY 11042. Received July 18, 1985; revision accepted Aug. 7, 1986. Vol. 74, No. 5, November 1986 from projected images of angiograms.7' Others have digitized angiograms and used the computer to determine the vessel edges.9' 10 Since most lesions are eccentric,'1 Brown et al.7 suggested that the diameter of a stenosis be determined for each of two orthogonal views and the area calculated by assuming that the lesion is an ellipse. However, some investigators believe that even two views may be inadequate and they advocate the use of as many angiographic projections as possible to assess the severity of a stenosis."' 13 In contrast, videodensitometry has been shown to accurately determine the relative stenosis for irregular lesions with the use of only one projection.12' 14 In the present study we evaluated a new method combining computerized edge detection and videodensitometry for quantitating absolute cross-sectional area with the use of only one projection. The accuracy of this method and the two-view edge detection method 1099 by gest on A ril 0, 2017 http://ciajournals.org/ D ow nladed from