想象动作中动态脑电的信息熵研究

想象动作可以引起脑电信号的特异性变化，如何通过提取脑电特征参数来识别想象动作的发生是目前脑-计算机接口（BCI）系统设计的技术关键之一。本研究使用两种信息熵（功率谱熵和小波熵）作为特征参数，对手部想象动作执行前后的动态脑电信号进行分析，并与目前常用的mu节律事件相关去同步（ERD）参数进行比较，以寻找特异性更强的动态参数。结果显示在想象动作对侧的运动感觉区域中，功率谱熵和小波熵均表现出与mu节律强度变化具有明显锁时关系的改变，且区分错误率得到明显降低。而在中顶部位置mu节律强度无明显变化处，熵参数对左右手想象动作的响应也具有明显的不同。结论，功率谱熵和小波熵具有良好的时域分辨能力和更准确的区分效果，可以作为特征参数用于在线BCI系统。

Study on Information Entropy of Ongoing EEG Signals during Motor Imaginary

Motor Imaginary can make specific influence upon the ongoing EEG signal.How to recognize this response from dynamic parameters is becoming one of the key techniques in brain-computer interface(BCI) system design.In order to obtain the dynamic parameters having better characteristic sensitivity,we have analyzed ongoing EEG signals before and after hand imaginary movements by two kinds of information entropy,power spectral entropy(PSE) and wavelet entropy(WE). Compared with event related desynchronization(ERD) coefficient in mu rhythm,we found that PSE and WE expressed time-locked variances on contrallateral primary sensorimotor area with obviously low error rate.Furthermore,on central area, although power spectral density of mu rhythm kept consistency,PSE and WE showed different responses to different hand act imagination.We concluded that PSE and WE could be used in online BCI system for their good temporal sensitivity and higher distinguish ability.