Spectral Graph Modeling of E/I Imbalance in Epilepsy

Overview

Drug-resistant focal epilepsy affects approximately 30% of patients with epilepsy, and accurate pre-surgical localization of the seizure onset zone (SOZ) remains a critical challenge. Current methods identify regions of abnormal electrical activity but provide limited mechanistic insight into why certain brain regions generate seizures.

This project applies the Spectral Graph Model (SGM) — a biophysically-grounded neural mass model — to resting-state MEG recordings to non-invasively characterize excitatory-inhibitory (E/I) dynamics at the regional level. By fitting the model to empirical power spectra, we estimate biophysical parameters reflecting excitatory and inhibitory coupling strengths, which we examine in relation to surgically-validated seizure onset zones.

The work aims to provide mechanistically interpretable, non-invasive biomarkers that complement existing pre-surgical evaluation methods — particularly for MRI-negative patients where conventional localization is most challenging.

Manuscript in preparation.