ANN based Classification of ECG Signals for Myocardial Infarction using Cross Correlation Concepts

Abstract

A standard clinical electrocardiogram signal plays a major role in preliminary screening of cardiac abnormalities. This work deals with classification of normal and inferior myocardial infarction and presents a method for artificial neural network based analysis of ECG patterns using cross correlation concepts and ECG feature analysis using a novel feature set. In this paper, five novel parameters are used as attributes for the ANN; they are maximum value and average value of cross correlation between normal template and sample to be tested, amplitude of Q wave peak, amplitude of S wave peak and QT zone amplitude sum. The ANN model trained using these attributes gives an accuracy of 100% on the selected training set and almost 87% accuracy on the selected test set.