Abstracts

Rationale: Surgical resection remains the best option for long term seizure control for patients with medically refractory focal epilepsy. Unfortunately, such treatment does not always result in seizure freedom, particularly in the long term. There is growing evidence that focal seizures may include an ictal network which may develop over time and potentially include more than one zone of seizure onset. Current presurgical evaluation applies structural and physiological imaging to identify the likely seizure onset zone. Improved techniques are required for identification of the complete ictal network both structurally and physiologically, and to establish the role of the primary seizure onset zone within the network. We demonstrate dynamic spatiotemporal changes in connectivity within the ictal network during temporal lobe seizures using a linear measure of synchrony, the Synchrony Index (SI). Methods: We analyzed intracranial EEG data from 11 seizures from a patient with confirmed hippocampal sclerosis who is seizure free 2 years after resective surgery. A previously reported analysis demonstrated that the visually determined electrodes of seizure onset could be objectively identified using parameters derived from our Synchrony Index. EEG was recorded at 200 samples per second. Using MATLAB, the SI was calculated for every possible electrode pair combination in non-overlapping, one-second bins over the duration of the extracted file. Videos were constructed to dynamically display the highest SI value connection for each electrode for each second beginning 100 seconds preictally and ending 150 seconds after seizure onset. Still images were generated to highlight specific time points. Results: Dynamic SI changes demonstrated clear spatiotemporal differences between the preictal, ictal and postictal time periods during 9 of 11 seizures. The preictal and postictal periods were dominated by diffuse, low SI value connections. During seizures, a nexus pattern repeatedly emerged with the electrode of visually determined seizure onset (SOZ) at the center with widespread connections. A large increase in SI value occurred first in the SOZ, then throughout the remainder of the electrodes. This was most persistent 10-20 seconds into seizures, which correlates well with our previously published data showing a large rise in SI value in the SOZ at that time. Overall, this pattern displays a network in which the majority of the recording electrodes are strongly functionally connected to the SOZ. Conclusions: Techniques for spatiotemporal display of seizure networks are needed to augment our understanding of the pathophysiology of epilepsy and our selection of surgical candidates. We have demonstrated the use of a linear measure of synchrony to display the emergent, dynamic network patterns during a seizure in a manner that objectively highlighted the SOZ. Further investigation is required to refine this approach and improve the analysis and description of this type of data.