Research on Removing Ocular Artifacts from Multi-Channel EEG signals

Abstract

Due to the fact that the traditional electroencephalogram (EEG) signal artifact filtering algorithm needs to synchronize refer Electrooculogram (EOG) signal and can not avoid losing some EEG information in the process of removing EOG artifacts, an ocular artifact filtering algorithm without refer EOG signals proposed. Firstly, EEG signals are decomposed into several independent components by fast independent component analysis (FastICA) algorithm, and the components of EOG are extracted according to fuzzy entropy value and threshold discriminant. Secondly, the components of EOG are separated into several intrinsic mode functions (IMFs) by complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) algorithm, and the noise-dominant IMFs are identified and denoised by adaptive threshold. Finally, the signals are reconstructed, and the performances of the algorithm are evaluated by relative root mean square error, correlation coefficient and power spectral density. The best results are obtained in the dataset published by klados and EEG motor movement/image dataset. The experimental results show that, compared with several competing algorithms, the proposed algorithm has better performance in removing EOG artifacts and retaining EEG information.