An empirical wavelet transform-based approach for motion artifact removal in electroencephalogram signals

Motion artifacts reduce the quality of information in the electroencephalogram (EEG) signals. In this study, we have developed an effective approach to mitigate the motion artifacts in EEG signals by using empirical wavelet transform (EWT) technique. Firstly, we decompose EEG signals into narrowband signals called intrinsic mode functions (IMFs). These IMFs are further processed to suppress the artifacts. In our first approach, principal component analysis (PCA) is employed to suppress the noise from these decomposed IMFs. In the second approach, the IMFs with noisy components are identified using the variance measure, which are then removed to obtain the artifact-suppressed EEG signal. Our experiments are conducted on a publicly available Physionet dataset of EEG signals to demonstrate the effectiveness of our approach in suppressing motion artifacts. More importantly, the IMF-variance-based approach has provided significantly better performance than the EWT-PCA based approach. Also, the IMF-variance based approach is computationally more efficient than the EWT-PCA based approach. Our proposed IMF-variance based approach achieved an average signal to noise ratio (𝛥SNR) of 28.26 dB and surpassed the existing methods developed for motion artifact removal.