Computer Search for Electrocardiographic Lead Directions to Optimize Diagnostic Differentiation

A computer was programmed to search for electrocardiographic lead directions, yielding optimal separations between records obtained from 500 normal subjects and 422 patients believed to have either left ventricular hypertrophy (LVH) or right ventricular hypertrophy (RVH). The majority of the patients with RVH (86%) exhibited chronic pulmonary disease with severe pulmonary insufficiency where development of chronic cor pulmonale was assumed. Autopsy confirmation of RVH or LVH was obtained in 38 cases. The diagnoses in the remainder were based on history and physical findings. Resolved leads derived from a corrected orthogonal system were tested for discrimination of the three groups. The number of false-positive and false-negative classifications was kept approximately equal. To simplify record analysis, the number of QRS measurements necessary for interpretation was limited to a single one. The optimal lead direction to separate normal records from those obtained in patients with LVH had an azimuth angle of 306° and an elevation angle of –58°. The lead for separation of cases of RVH was almost identical with horizontal lead X. Comparison between results obtained with optimized leads and 12-lead records showed approximately equal performance for LVH when sensitivity and specificity were kept the same. These results were based on up to eight measurements from 12 leads as compared to one measurement in a single lead. The diagnostic performance of optimized leads exceeded that of the 12-lead electrocardiogram when a set of the most commonly used criteria for LVH and RVH recognition was used. A simple analog device for obtaining optimal leads was designed which can be attached to conventional single-channel electrocardiographs. The feasibility of directional lead optimization by orthogonal lead resolution could be demonstrated with concomitant simplification of record analysis and greater ease in record taking.