Abstracts

Rationale: The accurate identification of the seizure onset zone (SOZ) and its regulatory networks is an essential step before a potentially successful intervention in surgical epilepsy. A large percentage of patients who undergo surgery (~20-60%) do not significantly benefit from it, highlighting our incomplete understanding of the network dynamics. In particular, how different brain regions and systems interact with one another in the interictal and preictal epochs, ultimately precipitating a seizure, is poorly understood. In this study, we investigate the behavior of the SOZ and non-SOZ regions in terms of the characteristics of their modular behavior over time. We apply a graph theory approach by detecting the functional modules in electrographic brain networks and we study the interactions between various brain regions. We analyze hours of intracranial, interictal EEG data recorded from patients with intractable epilepsy and study the spatiotemporal dynamic changes of each brain region's contribution to different modules. Elucidating these mechanisms can lead to a more precise targeting of the SOZ and its regulatory networks. Methods: Data Pre-processing: IEEG signals were recorded from 5 patients undergoing surgical epilepsy evaluation. Data from each epoch were divided into one second non-overlapping, weakly stationary time windows in accord with related studies. Coherence estimation: We constructed functional networks in each time window using multitaper coherence estimation, which defines a network connection between electrode pairs as the power spectral similarity of signal activity. We applied multitaper coherence estimation with time-bandwidth product of five and eight tapers in accord with related studies.Network Methods: A modularity maximization Louvain-like greedy algorithm was used to identify functional modules in the IEEG data. The intramodule strength z-score quantifies how well connected a node is to other nodes in its community. The partition coefficient measures how the connections a node makes are spread out between nodes in different communities. The recruitment corresponds to the average probability that a node is in the same community as other regions of its system. The flexibility of a node is defined as the number of times a node changes its community allegiance across network layers, normalized by the total number of possible such changes. The promiscuity of a node is defined as the fraction of all communities in which the node participates at least once, across all network layers. Results: We find that SOZ nodes consistently show higher integration with various brain modules other than they belong to, as measured by the participation coefficient. This highlights how highly connected SOZ nodes are to the rest of the brain and might explain the ability of these nodes to initiate and propagate seizures. Moreover, we find that SOZ nodes show a higher degree of recruitment compared to the rest of the brain. This implies that the SOZ nodes are more highly interconnected to its own system than to the rest of the brain. Importantly, this finding stands when we control for the spatial proximity of the SOZ system. We did not find differences in the nodal flexibilty, nodal promiscuity or intramodule degree between the SOZ and non-SOZ. Our findings show that the SOZ behaves differently versus other brain regions in terms of its dynamic functional connectivity and modular behavior. This could lead to easier identification and prediction of the SOZ, possibly only from interictal data without the need to capture seizures. Conclusions: 1-The SOZ nodes show higher integration with various network modules, significantly more than the rest of the brain. This possibly underscores the ability of the SOZ to initiate and propagate a seizure to distant brain regions and be influences by various inputs.2-The SOZ nodes show higher recruitment compared to the rest of the brain, even when we correct for physical proximity. This suggests we can identify the important regions to be targeted based on their interconnectivity patterns.3-These findings shed a light on the distinct network dynamics of the SOZ and could allow for the accurate identification of the SOZ only from interictal data. Funding: NIH/NRSA Ruth L. Kirchstein T32 Fellowship in Neuroengineering and MedicineInstitute of Translational Medicine and Therapeutics (ITMAT), Institutional Clinical and Translational Science Award (TL1)