Nonlinear neurodynamical features in an animal model of generalized epilepsy

Epileptic seizures result from the intermittent spatial and temporal summation of abnormally discharging neurons [3,7,20,24]. Complex dynamical interactions between brain regions act to recruit and entrain neurons by loss of inhibition and synchronization. Long-term continuous 4-channel intracranial electroencephalographic (EEG) recordings were obtained from a genetically engineered model of generalized epilepsy (n = 3) and littermate controls (n = 3) in order to perform nonlinear dynamical analyses of intracranial brain electrical activity. Signal processing techniques included reconstruction of the EEG signal as trajectories in a phase space, applying nonlinear indicators on the trajectories (i.e., short-term maximum Lyapunov exponent), and statistical index (T-index) for quantifying interactions among distant brain sites. Analysis of interictal (seizure-free) and seizure prone periods in 2 to 3 week old H218 epileptic knockout mice revealed that (1) the brain electrical activity is of higher order during the seizure-prone period, and (2) the interaction among brain sites is more active during the seizure-prone period. In addition, dynamical analyses do not show significant difference between interictal periods in epileptic mice and littermate controls. These results suggest that the development of seizures in an animal model of generalized epilepsy is determined in part by non-stochastic neural processes. Results farther suggest that it may be possible to identify the occurrence of seizures in advance through dynamical analytic distinction of interictal and seizure-prone periods.