Application of an EMG interference filtering method to dynamic ECGs based on an adaptive wavelet-Wiener filter and adaptive moving average filter

In dynamic ECG signals, EMG interference is a common type of noise and overlaps with the ECG signal spectrum. Herein, we propose an EMG-filtering method that combines an adaptive wavelet-Wiener filter and an adaptive moving average filter. First, the R-peaks detection method is used to divide the signal into individual heartbeats, and wavelet threshold processing is used to evaluate signal-to-noise ratio. The wavelet-Wiener filter's parameters are then set adaptively according to the SNR, and the resulting adaptive wavelet-Wiener filter is applied to each heartbeat. Next, the low frequency components between two heartbeats are windowed to calculate standard deviation, and the order of the moving average filter is set adaptively according to the estimated SNR and standard deviation. Finally, a moving average filter is applied to the low frequency components to filter residual EMG noise. Before and after applying the moving average filter, splicing and discarding are performed on both sides of the low frequency components, effectively preventing distortion at the transition between the high and low frequency components. We evaluated the proposed method's performance using different ECG databases, including arrhythmia and dynamic ECG signals with motion artifacts, and compared its performance with that of other methods. Experimental results show that the proposed method can significantly improve SNR and effectively preserve clinical information, especially in environments with severe noise. The proposed method utilizes the advantages of both applied techniques to reduce the noise in ECG signals with minimal distortion and can be used as an effective tool for denoising ECG signals.