Performance-based approach for movement artifact removal from electroencephalographic data recorded during locomotion

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Recording electroencephalographic (EEG) signals during human locomotion is crucial for understanding gait disturbances.
• Movement artifacts (MA) significantly contaminate EEG data due to electrode displacement caused by walking.
• Existing methods struggle to effectively remove MA in dynamic walking conditions.

Purpose of the Study:

• To introduce a systematic methodology for removing movement artifacts (MA) from electroencephalographic (EEG) signals acquired during treadmill (TM) and over-ground (OG) walking.
• To quantify the prevalence of MA across different locomotion settings.
• To provide a unified approach for MA removal in gait-related EEG studies.

Main Methods:

• Utilized a 32-channel EEG cap and a 3-axis accelerometer placed on the forehead during walking trials (TM and OG).
• Applied Independent Component Analysis (ICA) to separate EEG signals into independent components.
• Developed a novel method to identify and quantify MA using the participant's stepping frequency derived from accelerometer data.

Main Results:

• Observed increased physiological signals (e.g., neck EMG) with higher walking speeds.
• Identified artifact-independent components sharing spectral patterns with MA, peaking at stepping frequency.
• Demonstrated successful MA removal and data cleaning using newly established benchmarking metrics.

Conclusions:

• The proposed integrated methodology effectively removes movement artifacts from EEG data recorded during both TM and OG walking.
• The approach offers a unified solution for processing EEG signals in gait analysis.
• This method enhances the reliability of EEG data for studying neurological conditions affecting gait.