Skip to Main Content
During epilepsy seizure Electrocorticogram (ECoG) may change dramatically from a nearly chaotic signal (basal state) into a highly synchronized signal during a seizure, characterized by high amplitude and low frequency, and suddenly go back to the basal sate, making hard to identify them from time series. The epileptic seizure shows some stages as it is evolving, the here studied are: basal, preictal, ictal and posictal. As most of the bioelectrical signal, ECoG is a highly non periodical signal, so the most suitable techniques to analyze them are the Time-Frequency algorithms (T-F), allowing to follow up its frequency evolution through the seizure. Each seizure stage has a set of frequency components (atoms), showing up at different time. These components have their own particular characteristics, depending on the influence on time (duration) and frequency (bandwidth) they can be rated in three different classes: Rhythmic, Intermediate and Transitory. There are different amount of these components among stages, based on the density of this atoms, here we try to identify each of the epilepsy stages, decomposing short signal segments (epochs) into their atoms by means of the Matching Pursuit algorithm, an adaptive T-F algorithm. Signals were recorded intracranially from Wistar rats at the cortex level, seizures were elicited by kindling model before recording.
Date of Conference: 12-14 Nov. 2008