Localization Of Epileptogenic Foci Based On Video EEG Sparse Bayesian Learning

Epilepsy is a disease in which neurological and functional disorders occur in the local brain area or the entire brain caused by abnormal discharge of brain neurons.Globally,at least 65 million epilepsy patients suffer from epilepsy.The number of epilepsy patients in China is about 9 million,which has become China’s second largest neurological disease.Among them,30% of epilepsy patients need to be treated by surgical treatment,as their symptom can not be relieved by taking antiepileptic drugs.Accurate traceability of epileptic foci is the key prerequisite for successful surgical treatment.At present,clinically localization of epileptic foci is mainly performed by experienced neurologists and surgeons by combining clinical symptom,seizure period and inter-seizure EEG performance of patients during epilepsy,and other imaging examination methods such as f MRI.Thus,it usually takes a long time to make a basic judgment on the patient’s epileptic foci.So far,the common methods of localizing epileptic foci include the minimum norm(MN),standardized low-resolution brain electromagnetic tomography(s LORETA),etc.,but the positioning is not accurately.This article innovatively applies the method of sparse Bayesian learning(SBL)to the location of epilepsy focus,and builds a real head model based on each patient’s own MRI data,effectively improves the accuracy of traceability imaging.Based on this,the MATLAB software platform is developed for SBL-based location of epileptic foci.In this study,the total of 103 V-EEG data of 29 epilepsy patients come from the Department of Neurosurgery of Shenzhen Second People’s Hospital and Department of Neurosurgery of Shenzhen General Hospital.The data in the seizure period are sorted,cut,and each piece of data is extracted from 100 ms before the start time to 300 ms after the start(-100 ～ + 300 ms)to locate the epileptic focus.According to the different treatment methods,all patients were divided into two groups,surgical resection group and stereotactic EEG guided radiofrequency thermocoagulation group,the former 18 patients had 59 segments of V-EEG data,and the latter 11 people 44 segments of VEEG data.For each group of V-EEG data,the SBL traceability method was used to compare the results with the MN and s LORETA methods,also compared with the actual position of surgical resection or RF thermocoagulation of each patient clinically.Surgical recovery of each patient was rated by the International League Against Epilepsy at the follow-up.The traceability localization results obtained from all the data in this study are only consistent with the surgical resection location or radiofrequency thermocoagulation location,and the patient’s postoperative recovery rating is ILAEⅠ.The superiority of the SBL method is verified by comparing the CT image of the patient,the MRI 3D fusion result after the operation or the SEEG electrode 3D fusion result after the operation.In this study,the traceability location resultswhich obtained from SBL,MN,s LORETA and clinical V-EEG readings,are analyzed respectively at the data segment level and the individual patient level.The results show that the accuracy of the SBL method is significantly better than that of the MN and s LORETA methods,and there is no significant difference from the clinical results obtained by V-EEG reading.