Improving lesion conductivity estimate by means of EEG source localization sensitivity to model parameter

Area of Science:

• Neuroscience
• Biophysics
• Medical Imaging

Background:

• Electroencephalography (EEG) source localization estimates neural activity using scalp potentials.
• Accurate source reconstruction relies on precise conductive head models, which must account for brain lesions with differing conductivity.
• Uncertainties in assigning lesion conductivity values can compromise the accuracy of EEG source localization.

Purpose of the Study:

• To quantify the impact of uncertainty in brain lesion conductivity assignment on EEG dipole source localization accuracy.
• To evaluate the sensitivity of source localization to conductivity variations in different lesion types.
• To develop and validate a method for improving lesion conductivity estimation and subsequent source localization.

Main Methods:

• Utilized an eccentric-spheres head model with an eccentric bubble to simulate brain lesions.
• Simulated EEG measurements across 64 pathological scenarios.
• Performed inverse dipole fitting with lesion conductivity values ranging from 0.5 to 2 times the actual value.
• Proposed a residual error analysis method to refine lesion conductivity estimates.

Main Results:

• Incorrect conductivity assignment caused significant localization errors (up to 1.7 cm) for highly conductive lesions (e.g., liquid-filled).
• Localization accuracy was less sensitive to conductivity uncertainties for low-conductivity lesions (e.g., calcified tumors).
• The proposed residual error analysis method achieved lesion conductivity estimation with a few percent error.

Conclusions:

• Brain lesion conductivity is a critical factor influencing EEG source localization accuracy.
• The developed method effectively improves lesion conductivity estimation, leading to highly accurate source localization (<5 mm error).
• This approach enhances the reliability of EEG for mapping neural activity in the presence of brain lesions.