Denoising Electromyographic Signals via Stationary Wavelet Decomposition and Filtering

The measurement of electromyographic (EMG) activity may be used to diagnose neuromuscular disorders, to control electrical devices through muscle-computer interfaces, or to return motor function via neuroprosthetic devices. Unfortunately, EMG measurements will often contain noise, introduced by mechanical vibrations, electromagnetic radiation, or the electrical stimulation of tissue. We compare wavelet-based filtering to traditional Fourier-based filtering methods and demonstrate that wavelet-based approaches better conserve EMG signals in the frequency domain while removing noise measurable within the time domain.