Wavelet Based Empirical Approach to Mitigate the Effect of Motion Artifacts from EEG Signal

Abstract

Physiological signal such as Electroencephalographic (EEG) is often corrupted by artifacts during measurement and processing. These artifacts may corrupt the important topographies and signal information quality. The human health diagnosis needs a strong and feasible biomedical signal. Hence, the elimination of artifacts from physiological signal is a vital step. The Ensemble Empirical Mode Decomposition (EEMD) algorithm is used to convert input single channel EEG signal into a multi-channel EEG signal. This multi-channel EEG signal is further processed with Canonical Correlation Analysis (CCA) algorithm. Finally Discrete Wavelet Transform (DWT) is employed to remove the randomness available in the signal due to remaining artifacts. This technique is tested and evaluated against currently available artifact removal techniques using efficiency matrices such as Del Signal to Noise Ratio (DSNR), Lambda, Root Mean Square Error (RMSE) and Power Spectral Density (PSD) improvement. The improved parameters DSNR and by 28% and 17.81% respectively, pronounce the eligibility of the proposed algorithm to stand on top of currently employed algorithms.