Detecting myocardial infarction with a numerical expansion technique: effects of data pooling

The ability of the Karhunen-Loeve (K-L) numerical expansion technique to detect the presence of myocardial infarction is assessed by combining pools of QRS area data collected from 12 canines. First, potentials during normal sinus rhythm are recorded from 84 body surface electrodes. This set of potentials serves as control (CTL) data. Next, the dogs are divided into two groups for gathering data from both anterior and posterior infarcts in the left ventricle. One group consists of seven dogs all with anterior infarcts and a second group of five dogs all with posterior infarcts. One week after the infarcts are induced, potentials are again recorded during sinus rhythm. These potentials serve as anterior infraction (AI) and posterior infarction (PI) data sets. From the potential data, the QRS area for each lead is calculated for all three experimental populations. The CTL, AI, and PI QRS area data are grouped together in different combinations to study the effects of data pooling in identifying myocardial infarction with the K-L numerical expansion technique. The results indicate that the calculation of percent dipolarity of the QRS waveforms depends on the combination of data present in a given data pool