Abstracts

Rationale: A network approach to the analysis of intracranial recordings showed that the transition from interictal state to seizure is associated with changes in network connectivity and information flow between specific components of the epileptic network. Previous studies demonstrated that seizure-related changes in effective connectivity and network coupling are prominent in gamma frequency band. In this study, we have examined changes in effective connectivity during seizures emerging within epileptic network associated with focal cortical dysplasia (FCD). Methods: The 162 seizures from 21 patients (34±10 years) with epilepsy due to FCD were analyzed. To estimate the connectivity we used modified Directed Transfer Function (DTF) which provides high frequency (1 Hz) and temporal (1 s) resolution. Connectivity maps were constructed from band-pass filtered recordings which were segmented using one-second sliding window with 90% overlap. The connectivity variability derived from a preictal state determined the baseline for the detection of the changes. Nodes (EEG channels) which deviated from the baseline (p=0.05) were classified as SOZ nodes. These nodes were compared with the spatial profile of clinically determined SOZ. Results: Specific and highly significant connectivity changes between interictal and ictal state (p<10^-5) were observed. These changes were detectable in the frequency range 40-170 Hz and involved alteration of information flow within SOZ and flow directed from SOZ to surrounding network. Nodes which displayed significant fluctuations in connectivity (unstable nodes) co-localized with SOZ. To examine the predictive potential of unstable nodes to localize SOZ, we had performed a blind study when we estimated the average outgoing connectivity of each node. The best results were obtained for connectivity changes present during early stages of seizure mainly at six seconds after the seizure. At this time point, the sensitivity of the unstable nodes to mark SOZ was 79.6±29.3% (median 90%), and the positive predictive value was 55.6±14.1% (median 55.5%). Conclusions: A detailed assessment of the connectivity dynamics and identification of unstable nodes have diagnostic potential to mark the critical components of FCD-related epileptic network and increase the yield of presurgical examination. The modified DTF technique allows reliable separation of the networks components that actively generate seizures from the regions of passive seizure spread. Funding: Supported by grants Ministry of Health Czech Republic AZV 15-29835A, AZV 17-28427A, Czech Science Foundation 14-02634S and Student Grant Competition of CTU SGS15/198/OHK3/3T/13. Access to CESNET storage facilities provided under the programme eIGeR, CZ.1.05/3.2.00/08.0142, LM2010005.