Spectral Selective Canonical Correlation Analysis to Remove the Power Line Interference from HD-sEMG Signals

High density surface electromyography (HD-sEMG) is a technique to measure the spatial electrical distribution of muscle activity, and it is increasingly applied in fundamental neuroscience, biomechanics, and exercise physiology. The HD-sEMG signals encompass a wide frequency domain with both low and high amplitudes, making them more susceptible to various types of noise. One important source of noise is power line interference (PLI) which affects the signal at the frequencies of 50Hz and its harmonics (i.e 50Hz, 100Hz, 150Hz) and can mask the real value of the signal at these frequencies that contain important information about the muscle activity. Several methods have been proposed to remove this PLI. The Canonical Correlation Analysis method has proven its efficiency in HD-sEMG signal denoising compared to other techniques but still maintained residual PLI components. In this study, we propose a new variant of the CCA denoising technique to more effectively identify and eliminate all harmonics of the PLI while preserving the HD-sEMG signal originating from other frequencies.