Abstracts

Rationale: A network concept of focal epilepsy has been proposed where multiple interrelated brain regions contribute to the origin and maintenance of seizures.  The goal of this work is to contribute to efforts aimed at determining whether fMRI-based approaches to functional connectivity (FC) could identify such a network.  Brain networks mediating cognitive, sensory and motor functions have been found to be active even when the brain appears to be at rest.  These resting state networks (RSN) have been isolated out of the fMRI data using either a ROI-based approach or an independent component analysis (ICA) approach.  An actively investigated question is whether fMRI could identify pathological networks mediating epilepsy that may be operating interictally, distinct from functional RSNs.  Although activation networks of regions mediating and/or propagating epileptiform activity have been identified, it is uncertain if such regions interact within a network when epileptiform activity is not present.  Our hypothesis is that epilepsy-related networks are distinct from RSNs and could be derived from fMRI data.  Methods: We used fMRI analysis to construct FC maps of the SOZ, as determined from intracranial EEG (icEEG), in patients with focal non-lesional epilepsy undergoing a presurgical evaluation.  FC maps derived from the SOZ were obtained in epilepsy patients (n=10) and matched controls.  FC maps in patients were derived by placing a 5mm ROI over the location of the most active electrode within the SOZ and in the homologous region of a matched control.  These maps consist of all voxels whose BOLD time signal is significantly correlated with the time course of the SOZ ROI.  Separately, we performed ICA analysis, using FSL (FMRIB, Oxford), of the pre-processed fMRI data using the default automatic setting for ICA analysis.  ICA components were inspected, using the imaging analysis software Mango (UT Health, San Antonio) for overlap with the SOZ electrode, and the percent overlap with SOZ was calculated for each component.  Subsequently, a group analysis of % overlap was compared between the epilepsy and control groups.  We also visually inspected all ICA components to identify those components that appeared to match (look similar to) the map of the SOZ, pursuing the possibility that such components may be epilepsy-related. Results: ICA analysis of fMRI data was performed in 3 patients and 3 controls.  Automatic ICA analysis identified 13-15 components per subject.  56.41 % of the components in the patient group overlapped with the SOZ versus 55.5 % in the controls.  On visual inspection, a component matching that of the SOZ map was identified in one of the patients. Conclusions: 1. In this preliminary analysis of 3 patient-control pairs, there is no apparent increased representation of the SOZ region in ICA components.  We plan to extend this analysis to 10 patient-control pairs by the time of poster presentation.  2.  The finding of an ICA component matching the SOZ map may suggest the presence of a spontaneously occurring epilepsy network.  However, the SOZ map may represent a resting state network (RSN) that happens to also incorporate the SOZ region.  We are separately investigating the connection of the SOZ map to epilepsy by correlating the FC map with seizure propagation and with the extent of the irritative zone (Karkar et al., AES abstract 2016). Funding: Department of Neurology, UT Health San Antonio