Research On The Epilepsy Seizure Prediction Algorithm Based On Multi-channel EEG Analysis

Epilepsy is a chronic brain dysfunction syndrome caused by many diseases. It causes big pain to patients. Patients suffer from intractable epilepsy can't be treated sufficiently because of the finite treatment method. The early prediction of epilepsy seizure is crucial to study new treatment method and improve the quality of patients'lives. Many studies have been carried out aiming to find the new method to predict the seizure.Some early works on the predictability of seizures were limited to single channel analyses by technology insufficiency. These studies ignored the relationship between different channels and didn't consider the brain as a whole. With the advent of computers'storage capacity, epilepsy center can store large continuous EEG data. It leaded researchers to know more about the relationship between different channels and study multi-channel EEG for seizure prediction. Phase synchronization based on the Hilbert transform was measured by Mormann, and it received better results than previous works.Phase synchronization based on the Hilbert transform is measured for two time series and based on the new theory that phase synchronization decreases firstly and than increases before seizure. It provides a new method and theory to epilepsy research filed. But it had a big localization.This task focused on these problems and tried to study the seizure prediction algorithm, than put forward the phase synchronization based on the wavelet transform. The long encephalic EEG of 6 channels of 21 patients were analyzed, and R which denotes the phase synchronization between every two different channels can be calculated. When the R decreased in the preictal and than increased in ictal, we considered it as the epilepsy seizure is going to occur. The 57 seizures in the 86 seizures of 21 patients can be predicted correctly. After comparing the phase synchronization based on the wavelet transform with the phase synchronization based on the Hilbert transform and accumulated energy, we concluded that the phase synchronization based on the wavelet transform is a useful algorithm for seizure prediction.Clinical experiments were also carried out in the task. We used phase synchronization based on the wavelet transform to analyze 4 patients'EEG and the results also approved that phase synchronization based on the wavelet transform is applicable for seizure prediction.