Abstracts

We here apply, to our knowledge, the first formal analysis of the sequential stages of seizure dynamics. Our goal was to seek unique properties of the initial and termination phases of seizures, to better understand how seizures start and stop. We studied 24 seizures from 9 children: 12 scalp seizures from 5 children with partial complex epilepsy (selected as relatively artifact free from 79 consecutive records), and 12 intracranial records (from 16 consecutive records) from 4 children with a variety of seizure types and etiologies (gliosis, dysgenesis, mesial temporal sclerosis, and peritumoral). Work was performed with approval from CNMC and GMU Institutional Review Boards.
We developed a novel approach to multivariate linear discrimination of Fisher (1937). We measured 6 independent aspects of synchronization, using a variety of techniques to quantify amplitude and phase correlations between channels, within 1 second non-overlapping windows of EEG (23-64 channels). Careful statistical controls were used to guard against spurious correlations due to frequency content. We examined all possible partitions of these seizures into beginnings, middles, and ends, seeking the best separation and examining significance with both normal theory and bootstrap. We then examined the grand averaged results for common dynamical characteristics during seizure beginning, middle, and termination. Discrimination into 3 groups was clear for 11 of 12 intracranial seizures (chi-square p[lt]0.05 for 12/12 and bootstrap p[lt]0.02 for 11/12). Analysis of variance demonstrated that phase amplitude variances were significantly elevated during the middle of scalp seizures (df=59, F=7.39, p [lt] 0.0001), and during the initial period of intracranial seizures (df=59,F=3.4,p[lt]0.02), reflecting decreased phase synchrony (Figure). For both scalp and intracranial records, no consistent evidence of increased synchronization was evident in any seizure phase. We here report the first study of dynamical discrimination of seizure evolution. We found significant extraction of distinct initial and terminal phases from 23 of 24 scalp and intracranial recordings. No consistent evidence of increased synchronization was evident within any of these stages by any measure, consistent with recent intracellular findings (J Neurosci 22: 7297-07, 2002). Significantly decreased synchronization was evident within both scalp and intracranial seizures.[figure1] (Supported by NIH R02MH5006, K02MH01493)