A model-based objective evaluation of eye movement correction in EEG recordings

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Eye movement artifacts significantly contaminate electroencephalographic (EEG) signals, complicating analysis.
• Existing methods for artifact correction often face challenges with signal contamination.
• Objective comparison of different correction algorithms is crucial for advancing EEG analysis.

Purpose of the Study:

• To develop a quantitative and objective method for comparing eye movement artifact correction algorithms in simulated EEG signals.
• To evaluate the performance of six different artifact correction algorithms under various electrode configurations.
• To identify the optimal algorithm and electrooculographic (EOG) electrode setup for artifact correction.

Main Methods:

• Simulated EEG data were generated using a realistic head model and eye-tracker data, separating ocular and cerebral potentials.
• Six artifact correction algorithms were assessed using signal-to-noise ratio (SNR) before and after correction.
• Performance was evaluated across five EEG and four EOG electrode configurations.

Main Results:

• The second-order blind identification (sobi) algorithm demonstrated superior performance in artifact correction.
• Optimal performance was achieved when using 6 EOG electrodes in conjunction with the sobi algorithm.
• This combination proved effective across all tested EEG electrode configurations.

Conclusions:

• The proposed simulation method provides a robust framework for objectively evaluating EEG artifact correction techniques.
• The second-order blind identification algorithm with 6 EOG electrodes is recommended for effective eye movement artifact correction in EEG.
• This finding has implications for improving the accuracy and reliability of EEG-based research and diagnostics.