基于运动相关脑电特征的手运动方向识别

为了研究如何从无创运动相关脑电中提取运动信息作为上肢主动康复训练的控制命令,通过设计实验,使右手完成左、上、右3个方向的运动,同时采集脑电数据和右手运动信息.通过小波时频分析确认与右手运动相关的脑电频带,并提取其小波分解系数作为特征,采用支持向量机进行特征分类,根据方向识别准确率分析提取特征的有效性.结果表明,运动脑电delta和theta频段的小波系数特征可以有效区分右手不同方向的运动,方向识别准确率的均值接近65%,并且用准备阶段特征分类的结果普遍优于运动阶段特征,因此,在手运动之前诱发的脑电活动含有丰富的运动信息,可用于脑-机接口系统提取上肢主动康复训练的控制命令.

Recognition of Hand Movement Direction Based on Movement-Related EEG Characteristics

To extract effective motorial information from noninvasive movement-related EEG signal utilized as con-trol commands of active rehabilitation,an experiment was designed which the right hand moved in three directions respectively（left,top,right）and EEG data of reciprocating motion and hand movement trajectory were recorded. Firstly,wavelet analysis was applied to confirm the movement-related bands in frequency domain,and then wavelet decomposition coefficients were extracted as characteristics. Next,the support vector machine algorithm was selected and the effectiveness of feature extraction was estimated through recognition accuracy. The results demonstrate that, the wavelet coefficients of delta and theta bands of movement-related EEG used as characteristics can effectively dis-tinguish right hand movements in different directions and have nice classification accuracies,with the mean classifi-cation accuracy of subjects up to nearly 65%. Furthermore,the recognition accuracies adopting characteristics of pre-paratory stage are superior to that of motorial stage,indicating that EEG evoked by movement preparation has abun-dant movement information and can be used for extracting control commands of active rehabilitation of brain-computer interface（BCI）system.