A novel algorithm for removing artifacts from EEG data.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Traumatic brain injury (TBI) research increasingly explores electroencephalogram (EEG) for rehabilitation technologies like brain-computer interfaces (BCIs).
• Extracting movement-related high-gamma band signals from TBI patient EEG is difficult due to surface electromyogram (sEMG) artifacts from involuntary movements.

Purpose of the Study:

• To propose and validate a modified Independent Component Analysis (ICA) model for effective sEMG artifact removal in EEG data from TBI patients with hemicraniectomy.
• To assess the algorithm's ability to preserve crucial brain features, particularly the high-gamma band, post-artifact removal.

Main Methods:

• A modified ICA model was developed incorporating simulated EMG data added to raw EEG as extra channels.
• Independent components (ICs) associated with artifacts were automatically identified and rejected based on predefined criteria.
• The algorithm's efficacy was tested using EEG data from a healthy subject and a TBI patient with hemicraniectomy during hand movement tasks.

Main Results:

• The proposed algorithm successfully removed sEMG artifacts from EEG data by up to 86.72%.
• High-gamma band activity (80-160 Hz) was localized to the hemicraniectomy area post-artifact removal.
• The magnitude of gamma power during movement demonstrated improvement after sEMG artifact suppression.

Conclusions:

• The modified ICA model offers a robust solution for sEMG artifact removal in TBI patient EEG data.
• This technique facilitates the extraction of meaningful neural signals, crucial for advancing BCI and rehabilitation technologies for TBI survivors.
• The findings highlight the potential of high-gamma band analysis in understanding brain activity post-TBI and surgical intervention.