Combination time-frequency and empirical wavelet transform methods for removal of composite noise in EMG signals

Abstract

Electromyography (EMG) measures the electrical activity in muscles during movement. An electromyography machine records the electrical signals from the muscle’s movements as sound waves. EMG signal measurement is used to help doctors make a quick evaluation to identify conditions such as muscle problems; dystrophy and neuropathies. The objectives of this study were to provide a good method that gives proper preprocessing for EMG signals. Often, the EMG signal is corrupted by natural noise, which makes analysis difficult; the new method, namely the empirical wavelet transform, is used to remove this noise and make it clean. The non-stationarity of EMG signals makes the analysis more difficult by the conventional methods such as time domain and frequency domain, which do not give much information about the EMG signals. The time-frequency tool is unavoidable, in this research we used periodogram, Choi-Williams, and smoothed Pseudo Wigner-Ville. The obtained results of the denoising and time-frequency applications demonstrate the high performance of the Periodogram and EWT methods in comparing with other techniques previously published.