On the fully automatic construction of a realistic head model for EEG source localization

Area of Science:

• Medical Imaging
• Neuroscience
• Computational Modeling

Background:

• Accurate multi-tissue segmentation of MR images is crucial for realistic head conductivity models.
• Current segmentation methods for finite element head conductivity models (FEHCM) require manual input, which is time-consuming and subjective.
• This limits the efficiency and objectivity of electroencephalography (EEG) source localization.

Purpose of the Study:

• To develop and evaluate a fully automatic segmentation method for MR images.
• To assess the performance of the proposed method in constructing FEHCM for EEG source localization.
• To compare the automatic method against existing manual segmentation techniques.

Main Methods:

• A novel Hierarchical Segmentation Approach (HSA) was developed, integrating Bayesian-based Adaptive Mean-Shift (BAMS) segmentation.
• The HSA-BAMS method was evaluated using synthetic 2D and real 3D MRI head data.
• Segmentation and EEG source localization accuracy were compared against two reference methods and expert manual segmentation.

Main Results:

• The HSA-BAMS method demonstrated superior performance compared to the two reference methods in segmentation accuracy.
• The FEHCMs generated using HSA-BAMS resulted in accurate EEG source localization.
• The automatic HSA-BAMS method served as a reliable surrogate for manual segmentation.

Conclusions:

• The proposed HSA-BAMS method offers a fully automatic and accurate solution for multi-tissue MR image segmentation.
• This approach significantly improves the efficiency and objectivity of creating FEHCMs for EEG source localization.
• HSA-BAMS has the potential to replace manual segmentation in constructing realistic head models for neuroimaging research.