Vectorcardiographic identification of myocardial scar: a discriminative study with automatically processed...

A multivariate discriminative procedure for the vectorcardiographic identification of ischemic myocardial scarring was performed utilizing data from 1,162 vectorcardiograms (VCGs) obtained in clinically normal subjects and 90 VCGs obtained from patients proved at autopsy to have ischemic myocardial scars. The VCGs from patients with myocardial scars were divided into two groups, a design group of 50 cases and an evaluation group of 40 cases. The best vectorcardiographic variables to discriminate the clinically normal group from the design group with scars were identified by stepwise linear discrimination. Sixteen vectorcardiographic variables were then used for discriminative analysis. This analysis correctly identified myocardial scars in 45 of the 50 VCGs in the design group (sensitivity 90%); among the 1,162 VCGs from clinically normal subjects, 32 were misidentified as myocardial scar (specificity 97.2). The sensitivity of these defining criteria was then tested in the 40 cases of myocardial scar in the evaluation group and found to be reproducible; 34 of the 40 cases of this group (85.0%) were correctly identified as having a myocardial scar. The multivariate discriminative criteria developed in this study had greater sensitivity and specificity than standard methods usually employed in electrocardiography and vectorcardiography. The criteria defined need to be evaluated in a large series that includes instances of cardiac pathology of nonischemic nature.