Comparison of models for quantitative left ventricular wall motion analysis from two-dimensional echocardiograms during acute myocardial infarction

To develop quantitative analysis of regional left ventricular wall motion in the absence of a gold standard, an objective statistical measure to compare models of wall motion is described. This measure can be derived from wall motion analysis of subgroups of patients with different patterns of wall motion. A priori knowledge of the exact localization of wall motion abnormalities is not needed. Two-dimensional echocardiograms were analyzed from 79 patients with myocardial infarction. The following 4 models were compared: Model I was based on the descent of the base toward the stable apex during systole. Models II and III measured area reduction with fixed- and floating-reference systems, respectively. Model IV was the centerline model. Classification by the electrocardiogram of the myocardial infarction as anterior (n = 37), posterior (n = 17) and inferior (n = 25) provided the a priori probability for classification of myocardial infarction. The a posteriori probability for classification of myocardial infarction was derived from the detection of wall motion abnormalities by echocardiographic analysis. The mean difference between a posteriori and a priori probability is a measure for the diagnostic value of the model, and was measured for 200 regions/patient. Use of the described measure revealed model I to be the most informative model and model III the least informative. Thus, the described statistical measure contributes to the development of regional wall motion analysis