Automatic Processing of Cineventriculograms

To evaluate disorders of regional wall motion in patients with coronary artery disease, we developed a method to detect the boundary of the cineventriculograms with the aid of minicomputers. Each image of the left ventriculogram, taken at 60 frames/sec, was transferred to computer through a flying spot scanner and stored in the disc. The endocardial margin was automatically traced along the maximum value of spatial derivatives of the silhouette density using a new algorithm we developed. Each image was then superimposed using the external markers, and radial grids were drawn from the center of the gravity of the enddiastolic silhouette to the margin of the ventricular cavity. The length of grid lines was continuously measured throughout the cardiac cycle, which provided quantitative description of the segmental centripetal motion over the entire circumference. In 17 patients who complained of chest pain, single-plane left ventriculograms were obtained in both the resting state and immediately after cessation of rapid cardiac pacing. Patients with normal coronary arteries revealed synchronous wall motion in both states. In patients with narrowing of either one of the major branches, pacing stress increased the magnitude and extent of hypofunction or even induced abnormal wall motion in the ischemic area that showed normal contraction at rest, while there was slight enhancement of shortening in normally perfused area. In patients with three-vessel disease, wall motion over the entire ventricular surface, which was depressed before pacing, further deteriorated. Thus, the method we describe enabled precise quantification of the extent and severity of ischemic injury and of the functional status of the uninvolved myocardium. Pacing stress should be of considerable value in estimating coronary reserve.