Abstracts

RATIONALE: We have shown that epileptic seizures (clinical and subclinical) may serve as brain's neural resetting mechanisms (Epilepsia 1997; 38(S3):189) by reversing the observed long-term preictal spatio-temporal entrainment of normal brain sites with the epileptogenic focus. We recently produced further evidence along these lines by employing new methods for analysis of multichannel EEG recordings from patients with temporal and frontal lobe epilepsy. METHODS: Continuous, long-term EEG records (9 and 10 days) from 28 subdural and hippocampal electrodes in 2 patients with temporal lobe epilepsy were analyzed. Respectively, 19 and 24 seizures were contained in two records (patients 1 and 2). Measures of stability (Lyapunov exponents- Lmax) were estimated every 10 sec per electrode site. Convergence of Lmax profiles of selected sites over time is measured by a statistical measure (T-index). Integer quadratic optimization techniques were employed to select the optimal sites that minimize the average T-index estimated over these sites within a 10 minute period immediately prior to each clinical or subclinical seizure. The average T-indices of these sites were then estimated backward and forward in time. Preictal entrainment (T-index statistically zero) and postictal disentrainment (T-index greater than zero) at the ?=0.05 level were estimated with respect to the seizure onset time point. RESULTS: The range of preictal entrainment in patient 1 was [2.4 to 44.6] minutes with a mean (?) of 10.2 2.4 minutes. The range of postictal disentrainment was [1.7 to 48.0] minutes with ? of 11.0 2.9 minutes. The corresponding results in patient 2 for the preictal entrainment were [3.8, 127.5] and 27.3 5.7 minutes, and for the postictal disentrainment [0.5, 37.9] and 7.8 1.5 minutes. Therefore, long-term preictal entrainment of brain sites is followed by a seizure and, subsequently, by a postictal disentrainment. CONCLUSIONS: These findings may provide an answer to the question of why epileptic seizures occur and are related to the question of seizures' predictability, addressed earlier by our group (Epilepsia 1990; 31(S5):621, 1989; 30(S5)). This research is supported by NIH, VA and Whitaker research grants.