Ocular Artifact Removal Algorithm of Single Channel EEG Based on CEEMDAN-ICA-WTD

Abstract

Electroencephalogram (EEG) is one of the key technologies in the research of brain-computer interface (BCI). Removing ocular artifact (OA) from EEG signals is the premise of accurate and effective analysis of EEG signals. In the absence of reference electrooculogram (EOG) signals, the existing ocular artifact removal techniques will lead to distortion of the raw EEG signal during the process of removing OA. In this paper, an OA removal method based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Independent Component Analysis (ICA) is proposed, which is further combined with Wavelet Threshold Denoising (WTD). Different from other algorithms which directly zero the identified OA component, the identified EOG component is processed by wavelet threshold in this algorithm. The EOG component which contains EEG information is further separated to extract the pure EOG component. Results of these studies reveal that the CEEMDAN-ICA-WTD algorithm can remove OA signals while retaining the EEG signal to the maximum extent. Compared with the traditional single channel OA removal methods, the correlation coefficient between the raw EEG signal and the EEG signals which remove the OA can reach 97.9% for the proposed method. Furthermore, the root mean square error (RMSE) is also significantly reduced.