Global sensitivity of EEG source analysis to tissue conductivity uncertainties

Area of Science:

• Biomedical Engineering
• Neuroscience
• Medical Imaging

Background:

• Accurate electroencephalography (EEG) inverse solutions require precise head volume conductor models.
• Tissue conductivities within these models can vary significantly between individuals, yet this is often overlooked.

Purpose of the Study:

• To investigate the sensitivity of EEG forward and inverse solutions to uncertainties in tissue conductivities.
• To analyze the impact of these uncertainties on source localization across the entire cortex.

Main Methods:

• A detailed five-compartment head model (skin, skull, CSF, gray matter, white matter) was used.
• The finite element method (FEM) calculated EEG forward solutions.
• Generalized polynomial chaos (gPC) expansions generated numerous forward solutions efficiently.
• Goal function scans (GFS) were employed as the inverse approach.

Main Results:

• Skull conductivity showed the strongest influence on EEG signal topography for sources up to 4 cm deep.
• Skin conductivity had a notable effect, while white matter conductivity became more influential for deeper sources.
• Conductivity variations led to altered source localizations, especially for quasi-tangential sources.

Conclusions:

• Tissue conductivity uncertainties demonstrably influence EEG source analysis.
• Skull and skin conductivity variations pose the most significant challenges for accurate EEG source localization.