A Denoising Framework for ECG Signal Preprocessing

In this paper, Electrocardiographic (ECG) signal preprocessing is studied. The ECG signals from body surface are often contaminated by various kinds of noises such as power-line interference, baseline wander, electromyographic (EMG) noise, electrode motion artifacts and so on. These noises bring obstacle to the diagnosis of cardiovascular diseases. In order to eliminate the above noises in ECG signal, we have done a lot of experiments to suggest that the different de-noising algorithms to reject different types of noise. The combination of wavelet de-noising, band-pass filter using FFT filtering, and a nonlinear Bayesian filter is also introduced and demonstrated superior results compared with conventional ECG de-noising approaches. Finally, we apply this framework on the noisy ECG signals and show the excellent performance.