Computational system for the prediction of epileptic seizures through multi-sensorial information analysis

Abstract

During the past decades, several studies have demonstrated evidence that epileptic seizure may be preceded by variations in the patient’s brain dynamics. In this study, we introduce a Matlab toolbox that employs several mathematical methods based on advances presented by numerous investigation groups. The extracted features were processed for data visualization by space reduction (VISRED). The individual analysis of the extracted features did not present the expected results, although in several events, undeniable variations occur before seizures, suggesting that these variations could be related to pre-seizure activity. The combined analysis of all acquired variables, reveal more information comparing to an individual analysis. Data visualization by space reduction presented tenuous regions representing pre-ictal states. The progress of these ideas may assist the development of automated seizure warning devices for therapeutic purposes. Keywords:Epilepsy, Seizure Prediction, Brain dynamics, Data visualization