Multigroup diagnostic classification using body surface potential maps

Abstract

Multivariate analysis was performed on 120-lead electrocardiographic (ECG) data in order to derive diagnostic criteria for correct classification of 159 normal subjects (N), 103 patients with anterior myocardial infarction (AMI), 130 patients with inferior myocardial infarction (IMI), and 116 patients with pure left ventricular hypertrophy (LVH). The analysis used instantaneous voltage measurements obtained by sampling the time-normalized P, PR, QRS, and STT waveforms at equal intervals. The durations of these waveforms were measured prior to time normalization. Linear discriminant functions were computed for each possible bigroup comparison, and the six best discriminators of each pairwise comparison were selected for the final multigroup classification model. A total of eight features from five torso sites accounted for the correct assignment of 93% of N, 92% of AMI, 94% of IMI, and 82% of LVH. The misclassification matrix illustrated the relatively high rate of false negatives in LVH (11%). The improvement in classification over the standard 12 lead ECG was highest for LVH (11%) and lowest for N (4%); AMI and IMI rates were improved by 7% and 6%, respectively. >