[Quantitative, computer-aided determination of left ventricular contraction abnormalities in two-dimensional echocardiography. II. Application in patients with coronary heart disease].

In 105 selected patients with documented transmural myocardial infarction and segmental wall motion abnormalities in the 2-dimensional echocardiogram we determined the sensitivity, specificity and predictive accuracy of a semi-automatic computerized wall motion analysis method. The extent and localization of the LV-contraction abnormalities were documented by visual inspection of the 2-D echocardiograms by two experienced cardiologists and confirmed by angiographic and coronarographic findings. These findings were compared with computerized evaluation methods using an area-change method (5 areas) and a radial shortening method in a floating axis reference system. In the 2-chamber and 4-chamber views the area-change method was more sensitive (2-CH: 70% vs 49%, 4-CH: 79% vs 72%) and had a higher predictive accuracy (2-CH: 90% vs 87%, 4-CH: 86% vs 82%) than the radial shortening method. The specificity between the two methods was not different. In both views, however, we found distinct regional differences between the apex and the anterior and posterior wall. We conclude from our studies that in patients with myocardial infarction, computerized wall motion analysis in the 2- and 4-chamber views is accurate and reliable as compared to the visual diagnosis of two experienced observers. The area-change method in an intracardiac floating system seems to be superior to the radial shortening method.