Denoising ECG signal using different wavelet families and comparison with other techniques

Electrocardiogram (ECG) is a non-stationary biological signal. It detects the cardiac abnormalities by measuring the electrical activity generated in the heart. But ECG is very much sensitive. Its amplitude and duration can be corrupted by various types of noise, especially, power line interference, which sometimes leads to misdiagnosis. In this study different wavelet families identification and performance estimation is done for denoising ECG signal and the results are compared with adaptive NLMS and notch filter in both time and frequency domain. For evaluating the performance magnitude squared coherence (MSC), amplitude spectrum, power spectral density (PSD) and spectrogram are analyzed. SNR, %PRD, MSE, NMSE, RMSE, NRMSE and ESD performance parameter are also examined. Simulation is based on Signal Processing Toolbox built in MATLAB^®. The simulation result represent that wavelet transform is an excellent technique for denoising ECG signal.