Wavelet domain Wiener filtering for ECG denoising using improved signal estimate

A new two-stage algorithm for electrocardiographic (ECG) signal denoising has been proposed. It combines wavelet Shrinkage with Wiener filtering in translation-invariant wavelet domain. A time-frequency dependent thresholding has been proposed and grounded for obtaining a more adequate signal estimate in the first stage of the algorithm. It is related to ECC signal morphology and hence outperforms other thresholding approaches in this area. The experiments carried out on pathological and normal ECGs have shown better algorithm capabilities in comparison with other thresholding algorithms while suppressing parasite electromyographic (EMG) signals (the noise) and preserving diagnostically important ECG signal features.