Computerized approach for revealing coronary artery stenosis

Abstract

A computer-based approach has been used for finding significant leads and significant segments (P, QRS or ST) in each lead, for identifying coronary artery stenosis. The features for differentiation were derived from a singular value decomposition (SVD) of 12 lead digital electrocardiogram (ECG). The features denoted the position of the lead axes with respect to the first three components of the SVD decomposition. The methodology was utilized to differentiate a group of healthy individuals, patients with significant and non-significant artery stenosis (according to the angio-coronarographic findings), and three groups of patients with lesions in right, left anterior descending or left circumflex coronary artery. Segments giving the best separation together with values of sensitivity and specificity are presented.