Cortical graph smoothing: a novel method for exploiting DWI-derived anatomical brain connectivity to improve EEG...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Neuroscience

Background:

• Electroencephalography (EEG) source estimation infers brain activity from scalp potentials.
• This inverse problem is ill-posed due to high source dimensionality versus limited electrode data.
• Regularization is essential for stable and unique source estimation solutions.

Purpose of the Study:

• Introduce a novel regularization technique, cortical graph smoothing, for EEG source estimation.
• Leverage anatomical connectivity from diffusion-weighted imaging (DWI) to improve source localization.
• Evaluate the performance of cortical graph smoothing against established methods.

Main Methods:

• Developed a cortical graph smoothing regularization function based on brain's anatomical connectivity graph.
• Applied the method to event-related potential (ERP) data from simple motor tasks.
• Compared cortical graph smoothing against Minimum Norm (MN), Weighted Minimum Norm (wMN), LORETA, and sLORETA.

Main Results:

• Cortical graph smoothing demonstrated superior localization accuracy compared to the MN method.
• The novel method achieved greater relative peak intensity than MN, wMN, LORETA, and sLORETA.
• Evaluation was conducted across 18 subjects, validating the method's robustness.

Conclusions:

• Cortical graph smoothing offers an effective approach to EEG source estimation by integrating anatomical information.
• This method enhances the accuracy and resolution of brain activity localization.
• The findings suggest potential for improved analysis of neurological conditions and cognitive processes.