基于ECoG的运动想象脑-机接口分类方法

CLASSIFICATION OF BRAIN-COMPUTER INTERFACE OF MOTOR IMAGERY BASED ON ECoG

脑—机接口BCI(Brain-Computer Interface)技术是近年来国际上的研究热点之一,它通常利用脑电EEG(electroencephalo-gram)来实现无动作的人机交互,运动想象是其中一种重要的BCI实验范式。有关研究表明,脑皮层电位ECoG(electrocorticogram)具有更好的信噪比与频带特性。研究基于ECoG的运动想象BCI系统,针对ECoG信号的特点,改进了信号处理方法,提取数据的公共空间模式CSP(Common Spatial Pattern)特征,并利用支持向量机SVM(Support Vector Machines)进行分类器设计,提高了运动意向的识别正确率。用相应方法处理2005年脑-机接口竞赛中的一组实验数据,正确率达到92%,相比于当时参赛时所用的方法提高了6%。实验还发现,支持向量机在克服"维数灾难"和"过拟合"方面具有更好的鲁棒性。

Brain-computer interface （BCI） is one of the research hot point in the world recently. By using electroencephalogram, BC1 establishes no-work interaction between human and computer,in which motor imagery is an important experimental paradigm. Related studies indicate that Electroeortieogram （ECoG） has higher signal noise ratio （SNR） and broader bandwidth. BCI system of motor imagery （MI） based on ECoG is studied in this paper. Particular signal processing techniques are improved which exploits the character of ECoG information. By extracting the common spatial pattern （CSP） character of data and using support vector machines （SVM） to design the classifier, the recognition accuracy of MI is improved. Processing the dataset of Motor imagery from BCI Competition III in year 2005 by means of the above approaches, the result on testing set revealed an accuracy of 92% for classification, which was 6% higher than what we had submitted to the competition. Moreover, the result of experiment found that the SVM has better robustness in overcoming ＂curse of dimensionality＂ and ＂over-fitting＂.