运动相关脑电信号的运动意图预测方法研究

Research on Prediction of Movement Intention Method Based on Movement-related EEG Signal

摘要:

为了找出在大脑的后顶叶皮层区(PPC)运动意图预测与运动想象EEG信号之间的关联,联合运动相关电位MRPs与mu/beta节律的事件相关同步/去同步(ERS/ERD)特征,首先用小波包分解WPD重构特征频段的小波包分解系数特征向量,其次采用共空间模式CSP提取空域特征向量,最后利用支持向量机(SVM)进行运动意图预测。通过实验验证,联合运动想象信号中的运动相关电位及mu/beta节律,运动意图预测分类准确率达到85%。得出1)证实了运动相关MRPs可以表征运动准备即运动规划阶段的脑神经机制,2)10Hz以下的mu和beta节律ERS/ERD特征能够体现运动意图的方向。研究结论进一步为精细运动(包括运动方向、速度等其他运动参数)预测提供技术支持。

Abstract:

To find out how prediction of motor intention in the posterior parietal cortex(PPC) correlates with motor imagery EEG signal, this study joints movement-related potentials(MRPs) and the ERS/ERD features of mu/beta rhythm, in the first instance, wavelet packet decomposition(WPD) is proposed to reconstruct characteristic frequency band for feature vector of wavelet packet decomposition coefficients; moreover, spatial features vectors are extracted by common spatial patterns(CSP); in the end, support vector machine(SVM) as classifier is utilized to serve for predicting motor intention．Combining MRPs and mu/beta rhythm during motor imagery EEG signal, the classification accuracy is up to 85%. The result indicates that: 1) the brain nerve mechanism of movement readiness and movement planning stages can be characterized by MRPs; 2) the ERS/ERD features of mu/beta rhythm on low frequency components below 10Hz carry information about intended movement direction. And the conclusions further offer a technological support for predicting meticulous movement intention including direction, speed and so on of movement parameters.