Abstracts

Rationale: Dynamic causal modeling (DCM) (Neuroimage 2003; 19(4); 1273-1302) allows making inferences about effective connectivity underlying measured time series. When the approach was applied to intracranial EEG (iEEG) data for seizure activity, it could reveal modulations of synaptic connections responsible for the initiation of seizure activity (Neuroimage 2015; 107; 117-126). In this study we employed DCM to investigate dynamic changes in synaptic connections during the propagation as well as the initiation of seizure activity. Methods: We used iEEG seizure data recorded at 256 Hz sampling rate from intractable focal epilepsy patients at more than one-year follow-up period after epileptic surgery using a multichannel digital EEG system (Grass-Telefactor, West Warwick, RI, USA). For each seizure, we selected two sources, including the primary source (S1) which initially showed seizure activity from the seizure onset and the secondary source (S2) to which seizure activity propagated from S1 with a build-up pattern. From each of the sources, four segments of time series were extracted, which corresponded to the pre-ictal (before the seizure onset), early ictal (before the seizure propagation from S1 to S2), late ictal (after the seizure propagation from S1 to S2), and post-ictal (after the seizure offset) phases. Each time series segment with a duration of 20 s was divided into 2-s epochs using sliding time windows. We employed the canonical microcircuit-type neural mass model including extrinsic forward and backward connections between neuronal sources and intrinsic connections within neuronal sources in DCM for cross spectral densities. For each phase, we hypothesized 9 models of connectivity architectures (Figure 1) and identified the best model using Bayesian model comparison by pooling the evidence for the alternative models over the epochs. Results: The best connectivity architecture among neuronal sources changed during the pre-ictal, early ictal, late ictal, and post-ictal phases. Though not consistent between different seizures, the connectivity architectures with backward connections from S1 to S2 and forward connections from S2 to S1 (Models 4-6 in Figure 1) were identified to be the best during the early and late ictal phases, and the best connectivity architectures during the ictal period differed from those during the pre-ictal and post-ictal periods. Conclusions: During the initiation and propagation of seizure activity, extrinsic backward connections from the primary to the secondary ictal sources and extrinsic forward connections in the opposite direction were preserved, but the synaptic connections among the neuronal sources were differently modulated depending on phases during a seizure. Funding: Supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (2015R1C1A1A01052438 to CP and 2014R1A2A1A11052103 to HWL) and by the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI) funded by the Ministry of Health & Welfare (HI14C1989 to HWL).