Noise reduction based on ICA decomposition and wavelet transform for the extraction of motor unit action potentials.

We have studied methods for noise reduction of myoelectric signals and for extraction of motor unit action potentials from these signals. Effective MUAP peak detection is the first important step in EMG decomposition. We first combined independent component analysis and wavelet filtering to remove power line interference, and then applied a wavelet filtering method and threshold estimation calculated using wavelet transform to suppress background noise and Gaussian white noise. The technique was applied to single-channel, short-period real myoelectric signals from normal subjects and to artificially generated EMG recordings. In contrast to existing methods based on amplitude single-threshold filtering of the original myoelectric signal or a conventional digitally filtered signal, our technique is fast and robust. Moreover, the proposed algorithm is substantially automatic. The performance has been evaluated with a set of synthetic and experimentally recorded myoelectric signals. The basic tool for testing was power spectrum density (PSD) estimation by the Welch method, which allowed us to analyze the PSD of nonstationary signals.