ECG signal denoising using wavelet domain wiener filtering

A two-stage algorithm for suppression of electromyogram (EMG) artifacts from the electrocardiogram (ECG) using Wavelet Domain Wiener Filtering has been investigated. An improvement of the traditional technique is proposed by involving Time-frequency dependent threshold for calculation of the pilot estimate in the first stage. The appropriate choice of the wavelet basis functions used in each stage has been stressed. The strong relationship between the wavelet function´s support and the ECG morphology has been emphasized. The preliminary assumptions have been argued by experiments on a wide range database. They have shown that an appropriate choice of the decomposing wavelets for the two algorithm stages can considerably improve the quality of the denoised signal.