Computer Analysis of the Orthogonal Electrocardiogram and Vectorcardiogram in Mitral Stenosis

Abstract

Orthogonal electrocardiograms (Frank lead system) from 191 patients with pure mitral stenosis (MS) were compared to electrocardiograms from 510 normal patients and 296 patients with pulmonary emphysema (PE). All patients with MS underwent right heart catheterization; 121 patients underwent both left and right heart catheterization. Three hundred and thirty electrocardiographic measurements were obtained from each record and tested for their ability to distinguish mitral stenosis from normal. Two sets of diagnostic discriminators were selected: (1) ECG measurements which can be easily obtained by hand and (2) measurements which performed best in multivariate analysis using linear discriminant functions. The patients with mitral stenosis were separated into three hemodynamic subgroups according to mean pulmonary artery pressure (PAP) (0-30, 31-40, and > 40 mm Hg). The ability of the selected measurements to recognize mitral stenosis increased as mean pulmonary artery pressure rose. The superiority of the multivariate technique, using nine measurements simultaneously, was readily demonstrated when compared to the hand measurements. At a constant level of 6% false positives, multivariate analysis allowed a 74% recognition rate of mitral stenosis compared to 44% with hand measurements.