Electrocardiographic potential correlations: rationale and basis for lead selection and ECG estimation

Einthoven gave to us the electrocardiogram. Electrocardiographic mapping demonstrated that localized electrophysiological events and phenomena have localized body surface electrocardiographic manifestations. Clinical electrocardiography has given us "reasonably good" means (criteria) with which to detect and characterize medically significant cardiac conditions, events, and diseases. However, clinical electrocardiography is imperfect, largely as a consequence of inadequate or redundant spatial sampling. This compromises the sensitivity and specificity of diagnosing cardiac diseases for which electrocardiographic manifestations are present but undetected due to imperfect sampling that results in a low signal-to-noise ratio for the specific abnormality. Correlation structure of body-surface potential distributions between and across populations or individuals provides important insight into and justification for the selection and use of "limited", "reduced", or "derived" lead systems aimed at improving the capture and use of electrocardiographic information. In this paper, we show the electrocardiographic voltage correlation relationships that occur on the body surface, across groups of subjects, either with or without cardiac disease. In addition, we demonstrate the correlation relationships between torso-surface and epicardial-surface potential distributions in experiments incorporating isolated canine hearts in a human-shaped torso tank. Analysis of these correlation relationships provides an explanation for the long-standing success of clinical electrocardiography but also suggests the means to improve its performance by incorporating new leads and/or their estimation from appropriately selected leads.