Multi-compartment head modeling in EEG: Unstructured boundary-fitted tetra meshing with subcortical structures

Area of Science:

• Biomedical Engineering
• Computational Neuroscience

Background:

• Accurate human head models are crucial for electroencephalography (EEG) source localization.
• Existing methods often lack comprehensive discretization of subcortical brain structures.
• Modeling errors can significantly impact the sensitivity of non-invasive source localization.

Purpose of the Study:

• To introduce an automated, adaptable finite element (FE) mesh generation tool for multi-compartment human head models.
• To enhance the accuracy of EEG source localization by improving mesh quality.
• To address the lack of open software pipelines for discretizing complex brain structures in EEG studies.

Main Methods:

• Developed an automated approach using recursive solid angle labeling of surface segmentation.
• Incorporated mesh smoothing, refinement, inflation, and optimization procedures.
• Applied the technique to a magnetic resonance imaging (MRI)-based head segmentation including subcortical structures.

Main Results:

• Successfully produced an unstructured, boundary-fitted tetrahedral mesh with sub-one-millimeter fitting error.
• Achieved desired accuracy for 3D anatomical details, EEG lead field matrix, and source localization.
• The mesh generator was implemented in the open MATLAB-based Zeffiro Interface toolbox with GPU acceleration.

Conclusions:

• The automated mesh generation approach provides an accurate and adaptable tool for EEG studies.
• This method enhances the reliability of non-invasive source localization, particularly for subcortical regions.
• The integration into an open-source toolbox promotes wider accessibility and application in EEG research.