Separation of EOG artifacts from EEG signals using Hilbert-Huang transform

The electroencephalogram (EEG) signal is highly weak and usually contaminated by electrooculogram (EOG), this presents serious problems for EEG data interpretation and analysis. So, the automatic removal of EOG artifacts from EEG has been an important problem. In this paper, Hilbert-Huang transform (HHT) is applied to remove the EOG artifacts arising from eye movement. According to the local time-frequency properties of EOG and the statistic characteristics of intrinsic mode function (IMF) of raw EEG, the EOG contamination can be eliminated from EEG after threshold filter of IMF. The proposed method is fit for the non-stationary signal because of the highly perfect local time-frequency properties of HHT. The experiment results show that it is very efficient at automatically subtracting the eye movement artifacts.