Research On Response Mechanism And Decoding Technology Of EEG Induced By Compound Motor Imagery

Motor imagery(MI)is a kind of motion intention without any overt motor output.MI can result in event-related desynchronization(ERD)and event-related synchronization(ERS)of mu and beta rhythms over the sensorimotor area.The MIbased brain-computer interface(BCI)is the only active BCI that does not require external stimuli and reflects user's voluntary movement consciousness.It has played an important role in motor function compensation and brain plasticity induction.However,the current MI-BCI still faces some bottlenecks: 1)limited choices of the limb movement restrict the output of control instructions in an MI-BCI;2)the decoding performance of motion intention needs to be improved.Aiming at the above-mentioned key issues,this dissertation proposed compound MI induction strategies,including the compound limb paradigm involving multiple motor representations and the compound stimulation paradigm incorporating the exogenous signal,then developed related analyses and studies.For expanding the instruction set of an MI-BCI,two compound limb paradigms including simultaneous and sequential compound limb MI(CLMI)were proposed.Considering the cross-area activation pattern induced by simultaneous CLMI,three improved common spatial pattern(CSP)algorithms were designed.The results showed that the multi-class stationary Tikhonov regularized CSP(Multi-sTRCSP)achieved the best performance with the mean classification accuracy of ~70%,which verified the separability of seven mental tasks.Considering the time-variant EEG feature during sequential CLMI,two classification strategies based on Multi-CSP and power spectral density(PSD)were employed.The results showed that the PSD-based classification strategy was better and the mean classification accuracy of four tasks reached ~74%.At the meantime,the time scale could influence the effective recognition of sequential CLMI.The above results indicate that the simultaneous and sequential CLMI can effectively expand the command set of an MI-BCI and establish a novel multimodal MI-BCI paradigm.For improving the recognition performance of an MI-BCI,a novel compound MI-SSSEP paradigm was proposed by combining MI with steady-state somatosensory evoked potential(SSSEP)induced by electrical stimulation.Compared with the traditional MI-BCI paradigm,the classification accuracy of the novel compound MI-SSSEP paradigm was improved by ~14%,implying that the decoding performance of motion intention was significantly enhanced.It is proved that the compound induction paradigm combining MI and electrical stimulation is feasible and effective to enhance the performance of an MI-BCI system.In terms of the CLMI-induced brain response mechanism in the studies of decoding technology,significant differences in the ERD/ERS patterns and the activation range over the cerebral cortex were revealed between simultaneous CLMI and simple limb MI by investigating the EEG oscillatory pattern.And the source localization technology and the direct transfer function(DTF)were used to calculate the cortical EEG sources and construct the brain effective network,respectively.The sensorimotor areas corresponding to the involved limbs were found to be activated simultaneously during simultaneous CLMI,whose effective network showed more information interactions and dynamic connections between different functional areas.Furthermore,the prior sub-movement imagination can affect the neural activities of motor areas during the subsequent sub-movement imagination in the task of sequential CLMI.These results suggest that the cognitive motor process and information processing mechanism are more complicated during CLMI.The studies focusing on CLMI not only contribute to the understanding of brain function mechanism,but also provide a new theoretical basis for paradigm design and optimization of an MI-BCI.In conclusion,this dissertation developed multi-angle analyses and studies from the novel CLMI paradigm,the CLMI-induced EEG response mechanism and the novel compound MI-SSSEP paradigm,providing a scientific basis and technical support for multimodal and high-precision decoding of motion intention and deep exploration of complex cognitive function of the brain.