Research On Feature Extraction And Application Of EEG Based On Motor Imagination

With the improvement of economic level in recent years,the prevalence and mortality of various types of brain diseases have also increased year by year.These brain diseases bring different degrees of motor function loss to patients,causing great burden to patients and their families.Reduced people’s happiness index.Motor imagination is a special way of motor function state,not only does not depend on the residual motor function of the patient,but also can contact the active motion of the patient as a way to activate the motor neural network.Studies have suggested that motor imaging can locally activate a damaged motor network to improve the motor function of stroke patients.However,there is still no unified conclusion in the current research about the mechanism of motion imaging on the brain neural network.This paper focuses on the feature extraction and analysis of multi-class motor imaging tasks and brain rotation data of stroke patients,using traditional Granger causality,new causality methods and Lempel-Ziv(LZ)complexity.From the following three aspects:(1)A new type of motor imaging analysis method based on a new causality model is proposed to compare the differences in cerebral cortex activation under different tasks.Collected EEG data of 15 subjects under 8 tasks,and used the new causality model to calculate the causal relationship between channels under different tasks.Based on comparing the differences and connections between different tasks,it was found that KMI),VMI(Visual Motion Imagination)and Exe(Actual Motion)tasks have higher similarities in the brain area changes.And when doing visual tasks,the activation of the occipital region is more obvious,and the "cause" nodes are mostly located in this area..(2)A new brain effect network construction method is proposed to study the influence mechanism of different brain regions in stroke patients.EEG signals and behavioral data were obtained from three stroke patients in two periods(before and after training),and half-month’s mental rotation training was performed between the two periods.Firstly,LZ complexity and two causality methods were used for feature extraction and brain network construction of the EEG signal.The characteristics of the brain network under the motion imaging task were qualitatively analyzed based on the network laterality,the degree of the nodes,and the global network efficiency index.The statistical results show that the brain network connection has increased after training and the active areas of the brain have also expanded.The target of motor imaging in stroke rehabilitation is in the occipital region.(3)A method of integrating physiological information characteristics and behavioral data to evaluate the impact of motor imaging on stroke rehabilitation.To analyze the complexity of stroke EEG data and statistics of behavioral data,in order to avoid the subjectivity of EEG data and the contingency of behavioral data,a new index that combines the characteristics of physiological signals and behavioral datarehabilitation is proposed.,Used to assess the effectiveness of stroke rehabilitation.The greater the degree of rehabilitation,the better the effect of rehabilitation training.The results show that after a period of psychological rotation training,from the perspective of EEG data,behavioral data and fusion analysis,the three patients ’indicators have improved to a certain extent,and the behavioral analysis results show that the patients’ cognitive ability to the left hand is required Slightly weaker than the right hand,180 ° stimulus pictures are weaker than pictures at other angles(0 °,60 °,and 120 °).