Removal of ocular artifacts from the EEG: a comparison between time-domain regression method and adaptive filtering...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Ocular artifacts in electroencephalography (EEG) recordings can contaminate neural signals.
• Accurate removal of electrooculography (EOG) artifacts is crucial for reliable EEG analysis.
• Existing methods like time-domain regression have limitations in handling non-stationary EOG components.

Purpose of the Study:

• To investigate the performance of an adaptive filtering (AF) method for ocular artifact removal from EEG.
• To elucidate the roles of the forgetting factor (lambda) and filter length (M) in the AF method.
• To compare the accuracy of AF with the time-domain regression method under various conditions.

Main Methods:

• The study analyzes an adaptive filtering (AF) method for EEG artifact removal.
• The AF method's parameters, forgetting factor (lambda) and filter length (M), are examined.
• A simulation study quantitatively evaluates AF accuracy against the time-domain regression method.

Main Results:

• When lambda = M = 1, AF is equivalent to the time-domain regression method.
• A lambda < 1 accounts for non-stationary relationships between reference EOG and EEG EOG components.
• AF with M > 1 (e.g., 2 or 3) shows higher accuracy in recovering true EEG when EOG artifact shape differs or is misaligned with the reference EOG.

Conclusions:

• The adaptive filtering method offers improved accuracy for ocular artifact removal in EEG.
• The filter length (M) parameter is critical for enhancing accuracy when EOG artifacts exhibit shape differences or misalignment.
• AF provides a more robust approach for EEG signal processing in the presence of complex ocular artifacts.