Improved Somatotopic Consistency of EEG Source Localization Using a Personalized Segmentation-Free Head Model

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Somatosensory evoked potentials (SEPs) assess somatosensory pathways but face limited spatial resolution with EEG.
• Accurate source localization is crucial for understanding brain function and diagnosing neurological conditions.

Purpose of the Study:

• To investigate if a personalized segmentation-free head model and sparse inverse solver can achieve finger-wise somatotopy using EEG.
• To evaluate the model's performance against conventional methods and fMRI-based references.

Main Methods:

• Recorded SEPs from 16 healthy volunteers using a 65-channel EEG system during electrical stimulation of median, ulnar nerves, and individual fingers.
• Constructed personalized segmentation-free and conventional tissue-segmented head models from high-resolution MRIs.
• Computed forward fields using finite difference method and estimated P20/N20 sources with orthogonal matching pursuit.

Main Results:

• The segmentation-free model reduced localization differences by up to 2.9 mm compared to literature references.
• Source-location differences were comparable to or smaller than those from a standard FEM-based pipeline (10-15 mm).
• Demonstrated clear somatotopic progression from thumb to little finger stimulation.

Conclusions:

• Personalized segmentation-free head modeling enhances spatial consistency in EEG-based source localization for somatosensory targets.
• Forward model refinement, not intrinsic EEG resolution enhancement, contributes to reduced localization bias.
• Enables noninvasive assessment of somatotopic organization in the primary somatosensory cortex.