Performance analysis of ECG signal denoising methods in transform domain

Abstract

ECG signal denoising is one of the most critical step in any ECG signal processing task. This work provides a comparative study between two of the most widely used transform methods in ECG signal denoising problem. The first class of methods is the wavelet transform, particularly the discrete Wavelet Transform (DWT) and the Stationary Wavelet Transform (SWT). The second class of methods is the Empirical Mode Decomposition (EMD) and its variants namely Ensemble Empirical Mode Decomposition (EEMD) and Complete Ensemble Empirical Mode Decomposition with adaptive noise (CEEMDAN). This study is focused on the additive white gaussian noise (AWGN) considered as the most common source of noise generally studied in different ECG signal denoising algorithms. Simulations results tested on real ECG signals from MIT-BIH Arrhythmia database showed that the Stationary Wavelet Transform provides the better performance in terms of Signal to Noise Ratio (SNR), Root Mean Square Error (RMSE) and Percent Root Mean Square Difference (PRD).