Wavelet-basedwiener filter for electrocardiogram signal denoising

The aim of this study is suppression of parasite electromyographic (EMG) signals (myopotentials) included in ECG signals with use of the Wiener filtering in shift-invariant wavelet domain with pilot estimation of the signal. The wavelet filtering with hybrid thresholding was used for pilot estimation. The four-levels shift-invariant dyadic discrete-time wavelet transform decomposition was used for both main blocks of pilot estimation and Wiener filtering. Sampling frequency of used signals was 500 Hz. The testing set have included signals with small waves Q, high R waves and significant variations of precipitousness in onsets and offsets of QRS complexes. These signals were additionally noised by normal distribution noise its power spectrum was adjusted according to typical form of power spectrum of EMG signals