Wavelet Transform Based ECG Denoising Using Adaptive Thresholding

Abstract

Electrocardiogram (ECG) is a widely employed tool for the analysis of cardiac disorders and clean ECG is often desired for proper treatment of cardiac ailments. In the real scenario, ECG signals are usually corrupted with various noises during acquisition and transmission. As an important branch of wavelet transform, multiresolution has achieved good results in the noise reduction processing in many fields, such as ECG signal, voice signal, image signal and so on. However, multiresolution has strong dependence on the selection of wavelet threshold and wavelet function. In this paper, an adaptive wavelet threshold calculation and selection method is proposed. Based on the heuristic threshold optimization method, the adjustment factor of wavelet decomposition layer number and level influence is incorporated into the method. By dynamically adjusting the threshold calculation function for wavelet coefficients of each layer, more reasonable signal decomposition and noise reduction could be realized. The experimental results show that the proposed algorithm could achieve better performance in reducing the noise of ECG and could meet the needs of clinical application.