EMG Signal Enhancement Using Subband Soft-thresholding

Background: The electromyogram (EMG) signal indicates electrical activity of muscles, which is the summation of all motor unit action potentials within the detection area of the electrode. However, when the surface EMG signal is recorded, it inevitably contaminated with various artifacts originated from different sources like inherent noise in the electronic components of detection and recording equipment, ambient noise, motion artifacts, inherent instability of signal and other physiological signals. Although different methods have been proposed to denoising EMG signals the problem of accurate and effective de-noising technique of EMG still remains a challenge. Materials and Methods: In our study, a hybrid algorithm based on subband approach is implemented for EMG signal enhancement. Subband energy based enhancement using discrete wavelet transform (DWT) is first employed to separate high energy EMG component. Then the residual signal contains EMG and noise component. The EMG from the residual signal is extracted by using soft-thresholding. It is observed that wavelet thresholding works better for the signal with lower signal-to-noise-ratio (SNR), whereas, discrete cosine transform (DCT) based soft-thresholding performs well for higher SNR. A noise ratio factor (NRF) is introduced to select proper soft-thresholding method depending on the level of noise in the analyzing signal. Results: The proposed hybrid algorithm maximizes the performance. The EMG signals of healthy and myopathy patient are collected from publicly available dataset to evaluate the performance of different enhancement methods. The experimental results show that the proposed method performs better than the traditional soft-thresholding approaches for a wide range of noise levels. Conclusion: In this study, it is observed that DWT and DCT based soft-thresholding methods work better for low and relatively high SNR signals respectively. One decision factor is introduced to select the appropriate soft-thresholding algorithm between DWT and DCT. It improves the performance and hence the proposed method produces high SNR improvements of the EMG signal.