Electrocorticogram Based Causal Analysis Of Epileptogenic Networks And Epileptic Prediction

Epilepsy is a neurological disorder.As a result of seizures,patients will have a series of clinical manifestations.Some behavior will have some damage to patients with epilepsy,such as burns,drowning,fractures,accidents,death,etc.With the development of modern medicine,many epileptologists begin to study epilepsy and the methods for the treatment of epilepsy subsequently increase.Therefore,many patients with epilepsy can get treatment and return to a normal life.Although the development of modern medical science does not solve all the problems of the epileptic patient,people are filled with hope to take full control of epilepsy.Within the past decade,our understanding of the activity within the brain has switched from identification of disparate points of brain activity to identification of brain networks involved in information processing and task performance.A large number of studies have shown that the method of brain network can locaze the seizure onset.Therefore the localization for seizure onset zone(SOZ)based on epileptic network was studied,which verified the effectiveness of locating SOZ based on epileptic network.In addition,for some patients with refractory epilepsy,it is important to develop an accurate and effective epileptic prediction method.Because of the limitation of the database and the existing method of epileptic prediction,there has not been an effective algorithm which can be applied to clinical.Therefore,a method of epileptic prediction based on epileptic brain network was proposed,which can predict the upcoming epilepsy effectively.In this thesis,following the localization for seizure onset zone and epilepsy prediction,three parts of research works are conducted.Firstly,the theory of epileptic network was firstly analyzed,which can provide the basis for SOZ and epileptic prediction.Then,we studied the localization for SOZ based on epileptic network.The connectivity between the electrocorticogram(ECoG)signals was analyzed by the directional transfer function(DTF)method and the network properties were calculated by using the method of graph theory(degree,closseness and betweenness).Because the different centrality for implanted electrodes before and after seizures changes in this thesis,the data from three patients with clinical epilepsy were validated by this method.The results showed that the method was effective in locazing the SOZ.Then,based on the characteristics of epilepsy network before and after the onset,a method of epileptic prediction based on epileptic brain network was proposed.This method used support vector machine(SVM)to classify the three periods of seizure data from three clinical patients.The results showed that the classification accuracy rate is 94.82%,91.55% and 93.14%,respectively.Some moment can accurately trigger the alarm,and the time is 35 seconds,60 seconds and 85 seconds before the seizures onset.Then the method was applied to the data of four epilepsy patients in the database.The average predictive rates were 73.70%,88.32% and 80.54%,respectively.Through the above research,the obtained results can provide the basis and the direction for further study.Meanwhile,the algolothrim of epilepstic prediction proposed has good prospect in engineering applications.