Research On Active Closed-Loop Motor Function Rehabilitation System Based On Motor Imagery Brain-Computer Interface

Brain-computer interface is a kind of technology that enables the human brain to exchange information with devices directly.Brain-computer interface technology allows the human brain to immediately control external devices which are independent of peripheral nerves and muscles.Therefore,the brain-computer interface technology has been widely used in various fields,such as medical treatment,education,and military.On the background of motor imagery brain-computer interface,this thesis focuses on the research of the active closed-loop motor function rehabilitation system.Firstly,two improved motor intention recognition algorithms are proposed to meet the requirements of the rehabilitation system for its real-time performance and recognition accuracy.These algorithms are suitable for the case of few channels with low sampling rates and multi channels with high sampling rates,respectively.Secondly,an active closed-loop motor function rehabilitation system is proposed on the basis of algorithm research.Through the real-time electroencephalography processing framework research,a real-time electroencephalography processing system is designed.Then,based on virtual reality and mechanical hand rehabilitation robot,a paradigm of visual-auditory evocation and multi-sensory feedback is designed to improve the immersion of the subjects and the efficiency of rehabilitation training.Finally,the research work is verified by the experiments of open-source datasets and online acquisition.The specific contents are shown as follows:1.Two improved motor intention recognition algorithms are proposed for the online motor intention decoding problem.Under the conditions of low sampling rates or few channels,this thesis proposes an improved motor intention decoding algorithm.On the basis of traditional logarithmic band power and support vector machine algorithms,the improved algorithm divides the frequency band by performing a filter bank to get higher recognition accuracy.However,in the conditions of higher sampling rates or multi channels,the proposed improved logarithmic band power algorithm is difficult to meet the real-time requirement of the system.Therefore,this thesis proposes a motor intention decoding algorithm based on event-related potential and common spatial pattern.The proposed algorithm improves the recognition accuracy meanwhile reducing the computational complexity.2.An active closed-loop system structure is proposed for the design problem of motor function rehabilitation system.Firstly,this thesis designs an online processing system with Python,and standardizes the function of each module and communication protocols between modules.To ensure the processing speed,a real-time electroencephalography processing framework is designed based on multi-processing and lab streaming layer protocol.Then,the preprocessing,motor intention decoding,and feedback control are realized on this basis.Secondly,a multi-sensory evocation and feedback scheme are proposed based on virtual reality and mechanical hand rehabilitation robot.To improve the immersion and reality sense of subjects,two model training virtual scenes and four rehabilitation training virtual scenes are designed in this thesis.The scenes are designed according to hand flextion and extension training movements,which can improve the efficiency of the visual-auditory feedback.Then,a synchronous kinesthetic feedback method is realized based on mechanical hand rehabilitation robot,which expands the sensory modalities.3.Some experiments are designed and performed to verify the proposed motor intention decoding algorithms and the active closed-loop rehabilitation system.The proposed motor intention decoding algorithms are verified respectively by experiments based on open-source datasets and online acquisition experiments.It is proved that the algorithms can improve the online recognition accuracy compared with the traditional algorithms and meet the real-time requirements of the system.Besides,the designed active closed-loop motor function rehabilitation system is integrated,which is verified by the online acquisition experiment.It is proved that the system performs well both in evocation and feedback,and can effectively improve the efficiency of rehabilitation training.