ICA-based reduction of electromyogenic artifacts in EEG data: comparison with and without EMG data

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electromyogenic artifacts in electroencephalography (EEG) recordings during movement obscure brain activity due to overlapping frequencies and higher muscle signal amplitudes.
• Independent Component Analysis (ICA) is a common method for artifact reduction, but source separation can be imperfect.

Purpose of the Study:

• To evaluate the effectiveness of incorporating electromyography (EMG) signals as additional inputs to enhance ICA-based reduction of electromyogenic artifacts in EEG.
• To determine if adding EMG data improves the separability of independent components (ICs) representing myogenic artifacts.

Main Methods:

• Recorded simultaneous EEG and EMG data from nine volunteers performing seven distinct exercises designed to elicit myogenic artifacts.
• Applied ICA to EEG data, comparing results with and without the inclusion of EMG signals as additional inputs.
• Automatically classified ICs as myogenic or non-myogenic, removed myogenic ICs, and quantified artifact reduction using an objective measure.

Main Results:

• Including EMG signals as additional input to ICA significantly improved the reduction of electromyogenic artifacts compared to using EEG data alone.
• Objective measures confirmed enhanced artifact reduction when EMG data was incorporated into the ICA process.
• The study demonstrated that adding EMG data increases the separability of artifact-related ICs.

Conclusions:

• Incorporating EMG data into ICA is an effective strategy for improving electromyogenic artifact reduction in EEG.
• This enhanced artifact removal technique holds promise for research in locomotor disorders, brain-computer interfaces, and neurofeedback.
• The findings suggest broader applicability in analyzing brain activity during motor tasks.