Computer-assisted detection of ischemia utilizing echocardiograms before and after stress

Myocardial ischemia induced by stress may produce abnormalities of left ventricular (LV) wall motion. The location of wall motion abnormalities may be used to estimate the location and severity of coronary arterial obstructions. The American Society of Echocardiography has recommended the use of a 16-segment model of LV for assessment of wall motion abnormalities. Evaluation of motion of each segment of LV can be used to localize wall motion abnormalities (anterior, inferior, and lateral). Wall motion abnormalities can then be assigned to the distribution of each coronary artery. The normal response of the LV to stress is a uniform increase of regional wall motion, thickening and a reduction in end-systolic LV cavity volume, with minimal changes in diastolic cavity volume. Thus, echocardiographic comparison of LV performance before and after stress can be used to detect reversible myocardial ischemia. Computer-assisted detection of ischemia is based on a quantitative analysis of LV wall thickness at end-diastole and end-systole. This paper presents an approach to computer-assisted detection of ischemia utilizing proper boundary identification, segmentation and comparative analysis of wall segments. The proper boundary identification includes an image pre-processing stage, contour detection and segmental analysis. The algorithm used in this study combines the detection of endocardial and epicardial boundaries, and the computation of the area of segments of LV wall. The technique is a modified form of two-phase relaxation active contour detection. The analysis of the LV wall segments before and after stress includes the computation of the area of a segment at end-diastole and end-systole. These areas provide the key values for monitoring LV wall motion.