ECG Signal Analysis Using Principal Component Analysis

Abstract

Electrocardiogram (ECG) monitoring is the most important and efficient way of preventing heart attack and many other cardiac abnormalities. The analysis of ECG has become an important topic of medical research. Many algorithms have been presented in the literature for ECG signal analysis using different algorithms. In this work, we have proposed a method for Pattern Recognition and Classification of Atrial Premature Beat (APB), Left Bundle Branch Block Beat (LBBB), Paced Beat (PB), Right Bundle Branch Block Beat (RBBB) and Ventricular Premature Beat (VPB). We extract 450 signals from original database. We employ Principal Component Analysis (PCA) to extract the principal characteristics of the data. Then Classification Analysis is done using three best classifiers, namely, Support Vector Machines (SVM), k-Nearest Neighbour (k-NN) and the BayesNet with the Accuracy 99.34%, 99.34% and 98.54% respectively. This work gave better results as compared to other related work presented in the literature.