A comparison of adaptive and non-adaptive EEG source localization algorithms using a realistic head model

Area of Science:

• Neuroscience
• Biomedical Engineering
• Medical Imaging

Background:

• Accurate electroencephalogram (EEG) source localization is crucial for epilepsy surgery.
• The EEG inverse problem is underdetermined, requiring various algorithms with different assumptions.
• Existing algorithms include non-adaptive (minimum norm, LORETA) and adaptive (Borgiotti-Kaplan, eigenspace projection beamformers) methods.

Purpose of the Study:

• To compare the performance of four EEG source localization algorithms.
• To identify the most accurate and robust algorithm for clinical application.
• To investigate the impact of data window size on adaptive beamformer efficiency.

Main Methods:

• Investigated four EEG source localization algorithms: minimum norm, LORETA, Borgiotti-Kaplan, and eigenspace projection beamformers.
• Compared algorithm performance across various signal-to-noise ratio (SNR) values and single source locations.
• Assessed the influence of data window size on adaptive beamformer accuracy and efficiency.

Main Results:

• The eigenspace projection beamformer demonstrated superior localizing capabilities compared to the other three algorithms.
• This adaptive beamformer also minimized source current dispersion effectively.
• Optimal data window sizes were identified for adaptive beamformers, enhancing efficiency and reducing stationary source assumptions.

Conclusions:

• The eigenspace projection beamformer is a highly effective tool for EEG source localization in epilepsy.
• Its superior performance supports its use in surgical treatment planning for epilepsy patients.
• Further optimization of adaptive beamformer parameters can improve clinical applicability.