Phase space analysis of EEG in temporal lobe epilepsy

The electrocorticograms of human epileptic patients are analyzed by phase-space methods. The presence of an attractor is demonstrated as well as its low dimension during the preictal, ictal, and postictal period. The evaluation of the corresponding dimension of the phase space gives the minimum number of the necessary parameters. The integral correlation function provides insight into the dynamics of the neural networks involved in the three different stages of the phenomenon. The temporal change of the largest Lyapunov exponent is followed throughout the seizure giving a measure of the unpredictability inside the attractor. All of the above measures can be used for better mathematical modeling of the underlying process.<>