Limited Lead Selection for Estimation of Body Surface Potential Maps in Electrocardiography

Abstract

Body surface potential mapping has shown promise as a technique to improve the resolution and accuracy of diagnostic electrocardiography, but the cost and effort required to obtain maps have made wide spread use impractical. As a step toward a practical system, the problems of redundancy and uniqueness of electrocardiographic signal information contained in large numbers of leads were investigated. An algorithm for optimal selection of a limited number of leads was developed. Data obtained from 132 human subjects including some with normal electrocardiograms (ECG) as well as some with abnormal ECGs, were used in the study. Estimation of body surface potentials from limited leads was evaluated using three criteria, including rms error, mean correlation coefficient between limited lead and total lead maps, and error to signal power ratio. Using 30 leads the average rms error was 32 μV, average correlation coefficient was .983 and noise to signal power was 3.5% in the presence of 20 μV rms noise. Another finding was that optimal sites are not unique, i.e., different sets of optimal sites may be found which perform equally well. This result has practical implications for the design of lead systems for estimating maps on the critically ill and on patients undergoing stress tests.