Feature-based Stationary Wavelet Transform for Removal of EEG Ocular Artifacts

Ocular artifact is the prominent artifact in EEG signals that significantly affects the operation and interpretation of EEG-based clinical and control applications. This paper proposes an efficient ocular artifact identification and removal algorithm based on feature extraction and stationary wavelet transform (SWT). The EEG channels are decomposed into detail and approximation coefficients based on SWT. Three features, skewness, kurtosis, and peak amplitude of the approximation coefficient, are extracted and thresholded to identify ocular artifacts. The median filtering and inverse SWT are employed to remove the identified artifacts from the contaminated EEG signals. The proposed method is compared with the linear regression and DWT-based denoising techniques and evaluated on a publicly available dataset. To evaluate the effectiveness of the suggested approach, the performance measures “Signal-to-Noise Ratio” (SNR) and “Root Mean Square Error” (RMSE) are adopted. The $\triangle$ SNR values for the proposed algorithm and the other two methods are 26.03 dB, 15.87 dB, and 4.95 dB, respectively. The proposed approach significantly improves the signal quality of the reconstructed EEG signals.