An efficient wavelet analysis for ECG signal processing

Abstract

Fourier expansion, an excellent spectral decomposer of a signal suffers from a setback in its inability to represent the signal simultaneously in both time and spectral domain. ECG, a non-invasive recording method of bioelectric signal originated in the heart, provides valuable information about the electrical activity of human heart. Different features of the ECG can be extracted from the intervals and amplitudes of these waves at different sections. The primordial step in ECG feature extraction is the QRS Complex detection. The accuracy of the QRS complex detection defines the accuracy of locating all other waves and their intervals. The DWT, however a powerful tool for ECG feature extraction, is still limited by the choice of wavelets for use in feature extraction. We present a comparative study on the choice of some of the conventionally used wavelets and their corresponding accuracies for ECG feature extraction by R peak detection. In extracting, the B Spline wavelet transform of ECG signal is found to be the most efficient means of ECG signal processing.