| Epilepsy is a chronic brain disease caused by a variety of causes.It is characterized by repetitive,paroxysmal and transient central nervous system dysfunction caused by excessive discharge of brain neurons.Epilepsy has clinical manifestations of mental disorders,general convulsions,and loss of consciousness,which brings serious psychological and physical damage to patients.People of any age,region,and race will suffer from epilepsy,and its prevalence accounts for 1% of the world’s population,about 65 million people.Electroencephalogram(EEG)is the most important examination tool in the diagnosis,differential diagnosis and treatment effect evaluation of epilepsy.According to statistics,more than 80% of epilepsy patients’ EEG signals show abnormal discharge phenomena,including spikes,sharps,spike and slow waves,sharp and slow waves,and paroxysmal rhythm waves,among which the most typical one is spike discharge.This article mainly focuses on the analyzing and processing of epileptic EEG signals.Through identification and quantification of spikes in EEG signals,a method for automatic spike detection based on multi-channel EEG signals is proposed.The main research contents are as follows:1.A method of epileptic spike detection based on morphological filter and Random Forest(RF)model is studied.The proposed method uses morphological filtering technology to detect spikes from bipolar channels(BP)EEG signals,eliminates false positive spikes according to the threshold judgment method and obtain candidate spikes.Next the waveform features of candidate spikes are extracted to train the RF classification model.2.An epileptic spike detection method based on adaptive template matching and RF model is studied.The epileptic spikes generated by the abnormal discharge of EEG in different patients are slightly different in morphology.Traditional morphological filtering methods are difficult to accurately identify the epileptic spikes of different patients.On this basis,a candidate spike detection method based on adaptive template matching is proposed,in which the brain is first partitioned,the spike discharge area is found according to the signal energy.Second,the common template is used to identify the spikes on the average reference(AV)leads of this area,and then the spike results are clustered by K-means.After threshold screening,several clusters are obtained,and the centroid of each cluster is used as an adaptive template for the second template matching,and the result of the addition is the candidate spike detection result.According to the ‘peak-to-peak’phenomenon of spikes on the BP lead,the false positive spikes are eliminated,and then the morphological characteristic parameters of the candidate spikes are extracted,and the RF classification model is trained to realize the adaptive detection of epileptic spikes.3.Based on the above,an epileptic spike detection system based on multi-channel EEG signals are designed.At present,the existing EEG monitoring equipment in the hospital does not have the function of abnormal discharge detection.The identification and diagnosis of epileptic seizures is usually realized through the manual interpretation of spike discharges in EEG by doctors or EEG technicians.The whole process is time-consuming and labor-intensive.The accuracy of the diagnosis results depends on the personal diagnosis and treatment level of the doctor or EEG technician.At present,the existing professional medical personnel in our country are far from meeting the needs of artificial identification of epileptic spikes,and there is an urgent need for a system that can intelligently detect epileptic spikes to meet the needs of patients.In order to support theoretical research,this paper designs a set of epileptic spike intelligent detection system,which realizes collection and transmission of EEG data,and intelligent detection of epileptic spikes,which has been verified by the real EEG data of the Children’s Hospital,Zhejiang University School of Medicine. |
PDF Full Text Request