Classifying electrocardiogram peaks using new wavelet domain features

Abstract

We study distinctive properties of normal and malfunction electrocardiogram (ECG) peaks in the wavelet domain and based on this study we propose novel classification features for ECG signals. We analyze different combinations of the proposed wavelet domain and time domain features using multidimensional clustering and dimensionality reduction techniques. The results indicate encouraging accuracy rates.

Publication
COMPUTERS IN CARDIOLOGY 2008, VOLS 1 AND 2