Abstracts

Rationale: Different methods, from visual inspection of EEG recordings to complex computational techniques, have been used to localize seizures. However, as seizures evolve and spread to different regions of the brain during the ictal period, it may be difficult to either lateralize or localize their foci. In addition, propagation of electrical activity from the epileptic focus and consequent activation of distant neuronal networks remains partially understood. We, therefore, use modal analysis, a technique often used to solve inverse propagation problems, to transform intracranial EEG recordings of seizures to a time-mode space. We demonstrate that it is possible to localize the seizure focus, as well as accurately estimate initiation and termination of this interval in different brain regions, thus increasing our understanding of how epileptic events evolve and activate neuronal ensembles in otherwise normal brain regions. Methods: Intracranial EEG recordings from 5 adults with a history of complex partial seizures obtained during long-term (on average 10 days), monitoring were analyzed in this study. Recordings were obtained using several 6-8 lead subdural depth electrode arrays implanted bilaterally in 3 patients, and with a combination of subdural grids and strips implanted in 2 patients. At least three epileptic events were analyzed for each patient. The dominant modes, i.e, dominant components with distinct characteristic frequencies, of each EEG time series were extracted using mode decomposition and mode shape estimation techniques [Huang 1998, Stamoulis 1999]. The temporal and spatial distributions of peak modal amplitudes or mode participation factors were examined in the interval 5 min prior to the seizure onset to 5 min after the seizure termination. Results: Using modal analysis of intracranial EEG recordings we were able to accurately estimate the onset and termination of the ictal period in different parts of the brain and respective delays in signal arrival times associated with the propagation of the electrical activity from the seizure focus to the recording electrodes. In addition, we observed clear lateralization of the seizure during the ictal period, irrespective of the geometry and consequent spatial sampling limitations of the recording electrodes. Modal amplitudes were significantly higher in regions close to the focus of the seizure while activation of dominant modes was delayed and modal amplitudes were significantly lower in distant brain regions. Conclusions: Intracranial recordings provide a means of measuring synaptic activity associated with the activation of local neuronal networks due to either local sources or potentially distant but propagating events such as epilectic seizures. Our results suggest that during the ictal period of a seizure, local networks are differentially activated. However, it is still possible to identify the region of the seizure focus based on the distinct modal distribution and increased modal amplitudes of measured intracranial EEG signals in this region.