Discrete wavelet transform-based denoising method for real electrocardiogram

The electrocardiogram (ECG) is part of multi-component non-stationary signals. In the past, the ECG signal was studied in time or frequency domain independently. For that, the wavelet transform has been increasingly considered as a stronger time-frequency analysis and coding tool for those signals that might be contaminated by well known interferences such us the 50/60 Hz power line interference, the movement of electrodes and the breathing signal. In this work, we developed a Discrete Wavelet Transform (DWT)-based algorithm to denoise the ECG signals from the different aforementioned interferences. Then, we graphically investigate the different components of the ECG waves presented by their spectrograms, time-frequency images and scalograms. The developed tools are applied on both clean and contaminated ECG signals acquired from an ECG acquisition system, we have parallely designed. The experimental results show firstly the performance of our denoising algorithm to totally remove almost kind of interferences. Secondly, the graphical analysis of the ECG signals has given a new interpretation different from that known in time or frequency analysis separately.