Research On EEG Pattern Evoked By Single Hand’s Motor-imagery With Different Task Difficulties

Cerebral apoplexy also named stroke,was a disease of neurological deficits in brain evoked by cerebral vascular disease.About 80% to 90% patients suffered from physical disability of varying degrees after stroke.It was a great burden on society and family.Therefor,it was of great significance to carry out rehabilitation treatment for cerebral apoplexy patients.Research showed that repetitive moter imagery can promote the function of the brain area or the rebuilt the nerve pathway,it becomed a hot research topic in the rehabilitation treatment of stroke.Until now,which difficulty of imagining task can activate the brain area most was still unknown.It was of importance to study the EEG signal pattern based on the tasks of differential difficulty.According to the problem,two kinds of right hand moter imagery tasks were designed in this research,they were easy task and hard task.A total number of 21 subjects’ right hand’s grip signal and EEG signal were collected.The correlation analysis and error analysis were used to process the grip signal in the task of moter execution.The results showed that subjects under hard task do worse than easy task.Then the research processed the EEG signal evoked by moter imagery of different tasks.First,the research used the time frequency spectrum,power spectrum density and topographic map to analyse the EEG data of one subject.The result showed that the ERD(Event-related Desynchronization)phenomenon appeared in alpha band of 11-13 Hz and beta band of the 20-25 Hz,the ERD phenomenon in alpha band is the most significant and the duration is the longest.The statistical analysis of ERD values of all subjects showed that,the mean ERD of 21 subjects evoked by hard task is greater than ERD evoked by easy task.And ERD value of the two tasks in the beta band of 19-24 Hz had a significant difference.Finally,the research attempted to calculate the recognition accuracy of the two types of tasks.The EEG featers of two tasks were extracted by the CSP(Common Spatial Pattern)filters and classified by SVM(Support Vector Machine).Then the research optimized the classification algorithm by the frequency division method and time-frequency devision method.The results showed that the highest classification accuracy was generally in the beta band of 18-28 Hz,which meaned that EEG signal evoked by easy and hard tasks in beta band had the greatest discriminability.Finally,the research tried to combine the EEG signal evoked by moter imagery with the EEG signal evoked by moter executive to calculate the cross pattern recognition accuracy.The results of the whole study are expected to provide advices for the the rehabilitation of stroke when selecting the task to maximum activate the brain area.It could also increase the strength of EEG signal evoked by motor imagery for the patients who have the poor ability to finish moter imagery.