Research And Development Of Brain-Computer Interaction System For Motor Imagery Rehabilitation Training

Motor imagery(MI)training is a novel method of rehabilitation training.This paper explores the way to enhance the result of MI rehabilitation training,through the way of brain-computer interaction system based on neurofeedback.This paper proposed the system frame first.By researching on algorithms of articfact removal,feature extraction and pattern recognition,and programming to achieve the functions as the above algorithms,with the advantage of which,the system is to be completed.At last,this paper validates the effectiveness of the system through experiments.This paper can be summarized as follows:(1)Introducing the basic concept,compositions and research actuality of the system,as well as the key problems.Meanwhile,this paper intends to present the structure of brain,the mechanism of Motor Imagery Electroencephalogram(EEG)and ERD/ERS phenomenon to support the research.(2)Proposing the system frame and the function of each module should have.Designing EEG acquisition program,introducing the equipment,subjects and the points which should be payed attention to,and recording the situation of experiment are are three vital things to be employed.(3)Proposing a novel method for Ocular Artifacts(OA)removing automatically: Firstly,the horizontal and vertical Electrooculogram aliasing together,and together with EEG as the input to gain the independent components through the Improved Independent Component Analysis(IICA)algorithm.Secondly,use correlation coefficient to recognize and remove the aliasing signal component.At last,the EEG without OA is reconstructed using inverse transformation of ICA.(4)Feature extraction and classification of EEG can be divided into two aspects: obtain the wavelet energy as feature of EEG and constructing logistic classifier to classify characteristics.(5)At last,this paper completes system of Motor imagery rehabilitation training by programming.This system can not only analysis offline EEG,but also can analysis online EEG to achieve the classification.The results are to be turned into control signal which can control the virtual human body model and offer the feedback of situation subjects have.All in all,this system is helpful for MI training...