| Motor imagery(MI)is a mental process that only requires imagining the execution of a motor action but not physical movement.MI features independence and spontaneity not reliable on external stimuli and some other attributes,thus has become a focus of study in the fields of sports training,neural rehabilitation of stroke patients and brain-computer interface etc..Research on MI has made significant progress in feature information detection and machine learning decoding algorithms.However,there are still some problems such as the low overall recognition rate and the large differences in individual implementation effects,which put a bottleneck to the development of MI.To break the bottleneck,this article optimizes the quality of the motor imagination EEG sources(ie,the original signal)by means of “Guided Motion Video Inducement” and “Enhancing the Imagine task Difficulty”,and carries out both qualitative and quantitative analysis on EEG rhythm features based on event-related desynchronization(ERD)phenomenon.By quantitative indicators such as ERD mean value and recognition rate,the effect of optimization is assessed.The main work and results are as follows:First,the study optimized the EEG signal quality of motor imagery through the Guided motion video inducement.Comparison was made between video motor imagery(VMI)and non-video motor imagery(NVMI).The research results are as follows: VMI’s ERD was lower than NVMI(-2.9094<-1.7902,-2.3606<-1.6393),and ERD induced by VMI was significantly higher than that of NVMI.There was significant difference in ERD between the two task modes(P = 0.0002<0.01,R=0.755),which suggests that ERD intensity induced by VMI is significantly higher than that induced by NVMI.The std values for VMI were less than the std values for NVMI(1.2063<1.5598,0.9513<1.2038),the subjects exhibit a small difference in VMI and a large difference in ERD intensity for NVMI.In addition,the average recognition rate of VMI is higher than that of NVMI(81.37% > 71.63%).The discrimination recognition rate between VMI and NVMI was significant(p =0.0062<0.01,r=0.724).The above analysis shows that,the overall recognition rate in VMI mode is significantly higher than that in NVMI mode.Secondly,the study optimized the quality of motor imagery EEG source through increasing the difficulty of imaging tasks.This thesis carried out lower limb motion imagery research with different task difficulties,specifically high-frequency lower limb motion imagery(HMI)and low-frequency lower limb motion imagery(LMI).The results show that: The ERD means of HMI are lower than those of LMI(-1.827<-1.3487,-3.4756<-2.2891),and the ERD values of them show significant difference(p=0.0074<0.01,r=0.945).The RED intensity std values of HMI are less than those of LMI(0.4960<0.5279,0.6460<0.6725),which suggests that the ERD intensity difference among the subjects are small in HMI mode,and large in LMI mode.The average recognition rate of HMI is higher than that of LMI(87.84%>76.46%),with conspicuous difference existing between the recognition rates of the two modes(p=0.0034<0.01,r=0.429).In summary,the research of this thesis improved the quality of MI EEG source through the guided movement video inducement and enhancing the difficulty of the imaging task,so that the overall lower limb motion imagery recognition rate of subjects is improved and individual differences is reduced. |
PDF Full Text Request