Genetic particle filtering for denoising of ECG corrupted by muscle artifacts

Abstract

Suppressing electromyographic (EMG) noise in electrocardiogram (ECG) signals is a challenge, which shows frequently an impulsive nature and a wide spectral content overlapping that of the ECG. Most previous attempts of suppressing EMG signal are based on Gaussian noise modeling. This makes their methods susceptible to high-level EMG noise which is frequently coupled in the ECG signals under exercise conditions. To overcome this limitation, a new particle filter-based algorithm is develped for denoising of the non-Gaussian and non-linear ECG signals. Moreover, the genetic algorithm is used to mitigate the sample degeneracy of PF. Experiments show that our method could effectively suppress the EMG artifacts while preserving meaningful ECG components.