Date of Award

Fall 2016

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computational Analysis and Modeling

First Advisor

Leonidas Lasemidis

Abstract

Accurate epileptogenic focus localization is required prior to surgical resection of brain tissue for treatment of patients with intractable temporal lobe epilepsy, a clinical need that is partially fulfilled to date through a subjective, and at times inconclusive, evaluation of the recorded electroencephalogram (EEG). Using brain connectivity analysis, patterns of causal interactions between brain regions were derived from multichannel EEG of 127 seizures in nine patients with focal, temporal lobe epilepsy (TLE). The statistically significant directed interactions in the reconstructed brain networks were estimated from three second intracranial multi-electrode EEG segments using the Generalized Partial Directed Coherence (GPDC) and validated by surrogate data analysis. A set of centralities per network node were then estimated. Compared to extra-focal brain regions, regions located anatomically within the epileptogenic focus (focal regions) were found to be associated with enhanced inward directed centrality values at high frequencies (y band) during the initial segments of seizures (within nine seconds from seizures onset) and led to correct localization of the epileptogenic focus in all nine patients. Therefore, an immediate application of the employed novel network framework of analysis to intracranial EEG recordings may lead to a computerized, accurate and objective localization of the epileptogenic focus from ictal periods. The proposed framework could also pave the way for studies into network dynamics of the epileptogenic focus peri-ictally and interictally, which may have a significant impact on current automated seizure prediction and control applications.

Download

Share

COinS