Improved dipole localization using local mesh refinement of realistic head geometries: an EEG simulation study

Area of Science:

• Biomedical Engineering
• Computational Neuroscience
• Electrophysiology

Background:

• Accurate source localization is critical for interpreting electroencephalography (EEG) and magnetoencephalography (MEG) data.
• The boundary element method (BEM) is a common numerical technique for modeling EEG forward and inverse problems.
• Mesh density and dipole characteristics significantly influence the accuracy of BEM-based source localization.

Purpose of the Study:

• To systematically evaluate the dipole localization accuracy of the boundary element method (BEM).
• To investigate the impact of head model geometry, mesh density, dipole depth, and orientation on localization errors.
• To determine optimal meshing strategies and electrode configurations for improved EEG source localization.

Main Methods:

• EEG simulations were performed using both spherical and realistic head models.
• Dipoles were positioned in the right parietal and temporal regions with varying depths and orientations (radial and tangential).
• Uniformly meshed and locally refined models were analyzed, with potential data computed analytically or numerically.

Main Results:

• Localization errors of 2-4 mm were achieved with uniformly meshed models for dipoles deeper than 20 mm.
• Locally refined models were necessary for accurate localization of shallower dipoles (< 20 mm).
• Increased electrode count (19, 32, 63) significantly improved localization accuracy, particularly for shallow and tangential dipoles.

Conclusions:

• Dipole depth and orientation are critical factors affecting BEM localization accuracy.
• Local mesh refinement is essential for precise localization of superficial or tangentially oriented sources.
• Utilizing a higher density of EEG electrodes enhances source localization performance, especially in challenging configurations.