Abstracts

Rationale: Localization of seizure initiation and effective seizure control are the two sides of the same question. They require mapping the underlying network. However, the network underlying normal activity may not be the same as the one underlying seizure. The mathematics of structural controllability provides a clean formalism for classification of different types of functional connectivity with respect to control. Therefore we analyzed the functional connectivity derived from ECoG recordings during normal and seizure states, including interictal spikes and ictal activity.Methods: We computed the instantaneous phase of the oscillations filtered in theta, alpha and gamma frequency bands from intracranial recordings of patients with intractable epilepsy. Next we constructed the average phase profiles for each frequency band and state ([theta, alpha, gamma] x [normal, interictal, ictal]). We localized the sinks, sources, internal dilations and external dilations in the networks. We quantified the control profiles as ‘source dominated’, ‘external dilation dominated’ or ‘internal dilation dominated’. In an ANOVA general linear model design we compared the default, interictal and ictal networks with respect to the control profiles.Results: We analyzed intracranial recordings (n=32) from 8 patients. Our preliminary results suggest that: (1) sources vary according to the states and frequency bands of ECoG; (2) The source nodes of interictal spikes do not always overlap with the source nodes of ictal activity; (3) The control profile of interictal gamma activity is the inverse of normal activity (ANOVA n=32, P<0.01). The map of control profile (i.e., the source and sink positions) of ictal activity negatively correlated with the control profile of normal activity (n=8, Pearson’s r<-0.4, P<0.0001).Conclusions: Our study suggests that the control profiles of the network during seizure states differ significantly from the control profile of the normal (default) state and that there are significant differences between ictal and interictal spike network control profiles. The interictal seizure network control profile is the inverse of the default network control profile. This suggests that the activity flow in the cortical circuits is reversed during ictal activity: sinks of default network become seizure sources and sources of default networks become termination nodes (sinks). The implications of these findings are: 1) the inverse relationship between default network and seizure network enables prediction of seizure foci based on normal state ECoG recordings; and 2) the analysis of control profiles points to the potential targets of seizure control for implantable responsive neurostimulation devices.