The influence of electrode placement in the reconstruction and analysis of body surface potential maps from limited thoracic arrays.

Despite their capacity to indicate abnormality outside the scope of routine electrocardiography, body surface maps remain extensive, time-consuming research procedures. By contrast, a 35-electrode grid which sums precordial ST segment deviations has received wide attention as a clinical monitor of acute myocardial infarction. First, this study examined the feasibility of recovering essential data from a small electrode array to construct maps equal to those obtained from a much larger array. Such a small-array technique would offer economy, easy application, plus the comprehensiveness and clinical correlation of the large system. Second, the relationships between map, small-array and a 35-component equivalent multipolar generator were explored for a transformation system which both expands the small-array data to map displays and reduces such data to non-redundant waveforms. Comparisons were made between direct maps and those derived from two 35-electrode sets on normal subjects and patients with myocardial infarction or cardiomyopathy. Electrode placement did not conform to the conventional rectangular grid; for one, the electrodes encircled the thorax symmetrically; in the other they were statistically selected for signal information content. We found 1) symmetrical electrode placement and analytic reconstruction of maps from multipolar lead components consistently reproduced known maps well (.91 correlation, 120 microvolts error); but 2) empirical electrode placement and statistical prediction of known maps averaged .99 correlation and 20 microvolts error for the normal training population and .97 and 60 microvolts for the abnormal test sample. Worsening occurred when placement and prediction methods were mixed; however, maps reconstructed by the empirical-statistical approach reduced to a reasonable approximation of equivalent generator scalar leads.