The Recognition Of Motor Imagery EEG And Application In Upper Limb Rehablitation

Stroke is a kind of cerebral blood circulation disorders which is a serious threat to the health of elderly, about 3/4 of the patients suffer from limb movement disorder in different degree because of brain nerve damage. It’s not only bringing heavy mental pain and inconvenience of living, but also causing serious economic burden to the family and society. This is a social problem in the world. It has become an urgent social problem to help these patients to recovery and improve limb motor function, to increase their social participation ability and the ability to live independently in maximum possible. The problems are also puzzles of the modern rehabilitation medicine.The real motion and the motor imagery have the same motor neuron pathway and the similar relevance of the brain activation. Use brain computer interface technology to recognize the patients’ motor imagery electroencephalography(MI-EEG), it can help patients to translate their motion into control commands to drive the rehabilitation device to complete active rehabilitation training and improve motor function recovery effect. In this paper, the upper limb motor function recovery as the goal, the removal method of ocular artifacts in motor imagery electroencephalography, the adaptive pattern recognition method and the upper limb movement function rehabilitation system online controlled have been in-depth studied. The specific results are as follows:(1) The ocular artifact removal method based on LMD and CCABased on Local Mean Decomposition(LMD) and Canonical Correlation Analysis(CCA), an automatic method, denoted as LMDC, is proposed to remove the Ocular Artifact(OA) from electroencephalography(EEG). Each recorded EEG was decomposed into a series of physically meaningful production function(PF) components adaptively by LMD,and CCA was applied to eliminate the correlation among the PFs to get the corresponding canonical correlation variable. Then, the correlation coefficient matrix between each EEG and multi electrooculogram(EOG) was computed to recognize the OA component automatically. Furthermore, some random variables corresponding with OA components in the canonical correlation variable were set to zero, and the others remained unchanged to obtain a new canonical correlation variable. Finally, the inverse algorithm of CCA was utilized to project the new canonical correlation variable to the OA free PFs, and the OA removed EEG was reconstructed. Experimental research was conducted with Brain Computer Interface(BCI) completion database. Experiment results show that LMDC has better performance than the other common methods, and has stronger adaptability for multi subjects and types of OA.(2) An adaptive recognition method of EEG based on improved GHSOMTo enhance the generalization and adaptability of the recognition method for Motor Imagery Electroencephalography(MI-EEG), the Growing Hierarchical Self-Organizing Map neural network(GHSOM) was improved, and an adaptive recognition method based on Principal Component Analysis(PCA) and Improved GHSOM(IGHSOM) was proposed. The hierarchy growth judgment was automatically accomplished according to the quantization error of the expansion units in an upper layer. Thus, IGHSOM can reflect the mapping data in more details, and its stability and adaptive ability were improved. The experiment research was conducted to assess the recognition method, by which PCA was used to extract the MI-EEG features, and IGHSOM was employed to classify the features. The experiments results were presented on Brain Computer Interface Competition Data set, and the high recognition accuracy shows the correction and effectiveness of the improved strategy of GHSOM and the proposed recognition approach.(3) The online arm movement function rehabilitation system based on MI-BCIIn order to verify the validity of the method and to make MI-BCI limb function rehabilitation technology will be applied in the field of rehabilitation earlier, firstly, the EEG signal acquisition interface about imagination upper limb extension or flexion movement was designed based on the Qt cross platform of C++ graphical user interface application framework, and motor imagery real-time EEG data was recorded by combining with the C++ API function of g.MOBIlab+. Secondly, the ocular artifact removal method based on LMD and CCA and an adaptive recognition method of EEG based on improved GHSOM which were both proposed by this paper were used to process EEG based on MatlabR2010 a simulation software platform. Finally, the mechanical arm control system was built with the S3C2440 A ARM9 processor. EEG controlled mechanical arm extension or flexion on-line was completed based on mixed programming with Matlab and C++, in order to help upper limb movement dysfunction in patients to complete active rehabilitation training. The experimental results verify the feasibility of MI-EEG upper limb movement function rehabilitation system, and it shows its application prospect in the field of rehabilitation.The research results have a positive effect on the clinical process of promoting BCI technology in stroke patients with limb motor rehabilitation.