Performance Study of Different Denoising Methods for ECG Signals

Abstract

ECG is an important tool to measure health and disease detection. Due to many noise sources, this signal has to be denoised and presented in a clear waveform. Noise sources may consist of power line interference, external electromagnetic fields, random body movements or respiration. In this project, five common and important denoising methods are presented and applied on real ECG signals contaminated with different levels of noise. These algorithms are: discrete wavelet transform (universal and local thresholding), adaptive filters (LMS and RLS), and Savitzky-Golay filtering. Their denoising performances are implemented, compared and analyzed in a Matlab environment.