Leveraging network control theory and intracranial EEG for studying spectral brain state transitions in temporal lobe epilepsy

Epilepsy is a chronic neurological disorder marked by recurrent seizures. In drug-resistant focal epilepsy, effective surgical planning requires identifying the epileptogenic network, not only a single focus. This thesis proposes a multimodal pipeline combining stereotactic EEG (sEEG) with diffusion-MRI structural connectivity and Network Control Theory (NCT) to model seizure-related brain-state transitions. Spectral brain states were derived from band-limited sEEG power (β: 13–30 Hz; high-frequency: 80–500 Hz), mapped to an atlas whole-brain parcellation, and used as initial/final conditions of a constrained optimal control problem on each subject’s structural connectome. A geometric alignment and quality-control module matched sEEG contact coordinates to atlas space, enabling reliable contact-to-ROI assignment and an sEEG-informed partial control set. Within a continuous-time linear time-invariant framework, minimum-energy inputs were computed to drive the system from a pre-ictal baseline to an ictal state across frequency bands. Intracranial spectral analysis showed a broadband ictal power increase with frequency-specific spatial patterns: high-frequency activity was more focal and channel-selective, while β modulation was stronger but more distributed. NCT yielded similar total control energy across bands, but node-level energy rankings were stable in high-frequency ranges and partly reconfigured in β, suggesting complementary local vs network signatures of the transition. Although limited to a single case, the work demonstrates feasibility and motivates cohort-level validation and comparison with clinical targets and outcomes.