Effect of MRI Defacing on EEG Forward and Inverse Modeling

Area of Science:

• Neuroimaging
• Biophysics
• Medical Imaging

Background:

• Facial recognition improvements increase re-identification risk for structural neuroimaging (MRI).
• Defacing anonymizes MRI scans by removing facial features, but may affect subsequent analyses.
• Accurate electrical volume conductor models are crucial for electroencephalography (EEG) source localization.

Purpose of the Study:

• To quantify and map the effects of MRI defacing on individual head electrical volume conductor models.
• To assess the impact of defacing on EEG source localization accuracy using known ground-truth anatomy.
• To compare Boundary and Finite Element Modeling (B/FEM) approaches in the context of defaced MRI data.

Main Methods:

• Structural MRI scans from ten subjects were used.
• Individual electrical volume conductor models were created from both defaced and non-defaced MRI data.
• EEG source localization accuracy was evaluated using Boundary and Finite Element Modeling (B/FEM).

Main Results:

• Defacing structural MRI scans introduces errors in individual electrical volume conductor models.
• These errors lead to reduced accuracy in EEG source localization, particularly in specific head regions.
• The extent of mislocalization varies depending on the modeling approach (B/FEM).

Conclusions:

• Using defaced MRI data for EEG volume conductor model construction can significantly affect source localization accuracy.
• Clinicians and neuroscientists should be aware of potential localization errors when using defaced MRI data.
• It is recommended to build EEG volume conductor models from non-defaced MRI data prior to defacing to ensure accurate source localization.