Study Of Surgical Treatment Of Epilepsy Patients Based On Neural Computing Model

There are about fifty million patients with epilepsy in the world of which the number of Chinese epileptic patients accounts for about one fifth,and the number of patients with epilepsy is increasing year by year.Because of the paroxysmal and recurrent nature of epileptic seizures,they can cause irreversible damage to the brain function,and also have a serious impact on the quality of life and mental health of patients.Drug therapy is a routine and effective treatment,and surgical resection of epileptogenic zone is the first choice for patients with anti-epileptic drugs.Localization of epileptogenic zone based on the clinical experience of experts is the effective method,but some patients still have seizures after surgery.Therefore,it is important to predict the postoperative outcome by simulating the patient’s surgical resection before operation,which can help doctors optimize the operation plan and improve the probability of success.The research content of this paper is to predict postoperative outcomes in patients with epilepsy based on neural computational models.The connectivity matrix is used to correlate the nodes in the neural computing model,and an accurate connectivity matrix is helpful to improve the prediction result of the model.Therefore,an appropriate method is needed to determine the connectivity matrix.In this paper,a five-node network with known information flow among nodes is simulated and used as the input of the neural computing model,and then nine methods are used to analyze the relationship among the nodes.The connectivity matrix obtained by using nine methods is compared with the initial relational matrix.The simulation results show that with the directed transfer function algorithm using surrogate data,in the neural computing model,the connectivity relationship between nodes can be more accurately restored.Based on the determination of the electroencephalogram analysis method,the electrocorticography data of 12 patients with epilepsy during the interictal period were used to make the prediction.Firstly,the power spectrum method is used to select the frequency range of the electrocorticography energy concentration of epileptic patients as the frequency band for study and analysis.Secondly,the directed transfer function algorithm is used to construct the causality network of patients with epilepsy.Finally,surgical resection and random resection in patients with epilepsy were simulated based on the neural computing model,and then the postoperative outcome was predicted.The results showed that the prognosis of 10 patients with epilepsy was consistent with the actual outcome,and the prediction is accurate 83.3% of the time.The results show that the neural computing model based on the directed transfer function algorithm of surrogate data can accurately predict the postoperative outcome of patients with epilepsy in the frequency band of electrocorticography energy concentration.