Robust Electrocardiogram Beat Classification using Discrete Wavelet Transform

Abstract

This paper presents a robust technique for classification of six types of heart beats through ECG. Wavelet domain analysis is used for feature extraction from the ECG data along with instantaneous RR interval. Only 11 features are being used for this classification with a classification accuracy of ~99.5% through a 1-NN classifier. The main advantage of this method is its robustness to noise, which is illustrated in this paper through experimental results. Furthermore, Principal Component Analysis (PCA) has been used for feature reduction which reduces the dimensionality of the features from 11 to 6 while retaining the high classification accuracy. Due to its use of only a small number of features coupled with a simple classifier and its noise robustness, this method offers a substantial advantage over previous techniques for implementation in a practical ECG analyzer.