Phase-space reconstruction of ECoG time sequences and extraction of nonlinear characteristic quantities

The attractors, obtained from the ECoG time series of anaesthetized SD rat before and after epileptic seizure, are reconstructed first by making use of time-delay coordinates. Many efficient approaches and analysis techniques are applied to the ECoG series, thus the ECoG series are exactly analyzed. Consequently, a constructive result is obtained. Through calculating mutual information function, its first local minimum is defined as the time delay. And the method of uniting false nearest neighbour with Cao method is used to determine the optimal minimum embedding dimension. And then the ECoG sequences are considered as a chaotic series combining nonlinear prediction with surrogate data method, and are educed to be not a low-dimensional chaos. The largest Lyapunov exponents are computed on the basis of phase-space reconstruction of ECoG, at the same time the approximate entropies are calculated. The computational results show that there are distinct differences in the largest Lyapunov exponents and the approximate entropies before and after epileptic onset, which can provide with some clues for understanding the mechanism of epilepsy and predicting epileptic seizure and curing epileptic patients.