Hybrid MVMD-ICA Framework for EOG Artifacts Removal from Few-Channel EEG Signals

During the acquisition of electroencephalogram (EEG) signals using the portable few-channel EEG device, the recorded signals are susceptible to interference from electrooculogram (EOG) artifacts. This paper proposes an EOG artifacts removal method combining multivariate variational mode decomposition (MVMD) with independent component analysis (ICA). MVMD is applied to decompose the original few-channel EEG signals into multiple intrinsic mode functions (IMFs). ICA is used to separate these IMFs into independent components (ICs). EOG artifact components are removed through the sample entropy (SampEn) threshold standard, followed by the reconstruction of the EEG signals. Compared with other algorithms, the proposed MVMD-ICA framework efficiently eliminates EOG artifacts while preserving original EEG information, which is suitable for few-channel EEG signal preprocessing.