Analyze Temporal Lobe Epileptic Network Using Stereo-Electroencephalography

Epilepsy is a disease in which the brain produces abnormal discharges that cause abnormal brain function.Seizures are often accompanied by a variety of complication symptoms,which pose a great threat to the patient’s health.Taking inhibitory drugs all year round will bring heavy mental and economic pressure on the family.About 30% of patients with epilepsy are unable to control seizures with drugs and develop drug-refractory epilepsy.Temporal lobe epilepsy is the most common type of epilepsy among all patients with epilepsy.The methods of treating epilepsy mainly include drug control and surgical treatment,in which the identification of the lesion and the surgical removal of the epilepsy focus are important methods for the treatment of refractory epilepsy.Stereo-electroencephalography(SEEG)is a invasive recording method of EEG activity.This method can avoid craniotomy surgery,which results in smaller wounds and flexible electrode placement.It has been widely used today For the location of clinical epilepsy lesions.In terms of the distribution of epileptic brain regions,the epilepsy network has become an important means to locate epileptic regions and effectively treat epilepsy.In this paper,the clinically collected SEEG data are processed with bipolar leads,and an isolated effective coherence(iCoh)network model of temporal lobe epilepsy is constructed.The EEG of the epilepsy network at the early,middle and late stages of the seizure are calculated respectively.δ,θ,α,β,and γ,the in-degree,out-degree,and betweenness centrality(BC)network characteristics of the five frequency bands,and further uses the K-means method to analyze the epileptic clustering of the SEEG channel,and Comparison of the patient’s surgical area to determine classification accuracy.In order to visualize the analysis results,the patient’s MRI and CT images are also fused to realize the three-dimensional visualization of the positioning area.The analysis results show that using the bipolar lead processing method on the SEEG data can obtain higher quality local electrical activities and reduce interference.The SEEG epilepsy network constructed based on the iCoh method has a high classification accuracy in the δ band output value.Based on determining whether the patient is temporal lobe epilepsy,it can also accurately distinguish the position of the temporal lobe epilepsy channel.The BC value shows a higher classification accuracy in the γ band.This thesis further studies the network that combines the SEEG network and mathematical calculation models,and analyzes the construction of methods such as directed transfer function(DTF),partial directed coherence(PDC),iCoh and Pearson correlation.The results show that the PDC network combined with the mathematical calculation model can better locate the epilepsy area.The above research results show that the network structure analysis of SEEG signals in patients with temporal lobe epilepsy can accurately extract the characteristics of epilepsy areas,and provide a quantitative reference for clinically positioning patients with epilepsy areas.