Performance of beamformers on EEG source reconstruction

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• New beamformers for electroencephalography (EEG) and magnetoencephalography (MEG) source reconstruction are emerging.
• Comparative performance data for these novel beamformers are scarce.
• Understanding beamformer efficacy is crucial for accurate neural activity analysis.

Purpose of the Study:

• To evaluate and compare the performance of eight scalar beamformers for EEG neural source reconstruction.
• To assess beamformer effectiveness under various parameters and background noise conditions.
• To identify optimal beamformers for dipole time course reconstruction from EEG data.

Main Methods:

• Simulated EEG signals were generated using forward head modeling.
• Artificial dipoles were projected onto scalp electrodes.
• Simulated signals were superimposed on background activity, including real EEG and white noise.
• The performance of eight scalar beamformers was systematically analyzed.

Main Results:

• The eigenspace beamformer showed improved performance for small and large dipoles but is not suitable for real-time applications due to automation limitations.
• Minimum variance and Borgiotti-Kaplan beamformers exhibited the best overall performance.
• These top-performing beamformers demonstrated robustness to parameter variations and ease of application.

Conclusions:

• Minimum variance and Borgiotti-Kaplan beamformers are recommended for EEG neural source reconstruction.
• Beamformer selection should consider performance, robustness, and practical implementation.
• Further research may explore real-time capabilities and advanced beamforming techniques for EEG analysis.