基于想象左右手运动脑机接口实验研究及分析

Analysis and Research of Brain-computer Interface Experiments for Imaging Left-right Hands Movement

探索一种实用的基于想象运动思维脑电的脑．机接口（Brain．computerinterface，BCI）方式，通过寻找合适的信号处理方法，来提取最能反映不同思维的脑电特征，以提BCI系统通讯识别正确率，为最终实现BCI应用奠定理论和实验基础。对6名健康受试者进行3种不同时段（箭头出现2s、1s和0s后提示按键）情况下想象左右手运动思维作业的信号采集实验，利用小波分析、前向反馈神经网络（BP神经网络）对离线实验数据进行处理和分析。对所有受试者三种情况下的延缓时间△t2、△t1和At0分析发现：At0与△t1和△t2之间都有显著性差异（P〈0．05），而△t1与△t2之间没有显著性差异（P〉0．05）；三种情况下，平均分类正确率分别达到65．00％、86．67％和72．00％，实际按键前0．5～1s左右，想象左右手运动的思维脑电特征信号都发生明显改变，且这些特征存在明显不同。在箭头出现1s左右后提示随机按键情况下，可以获得更高的识别正确率，说明该方案提取的特征作为BCI系统外部装置控制信号是可行的，通过合理的实验设计获取的信号有助于识别正确率的提高，为BCI系统中思维任务的特征提取与识别分类提供新思路和方法。

This is a research carried out to explore a pragmatic way of BCI based imaging movement, i.e. to extract the feature of EEG for reflecting different thinking by searching suitable methods of signal extraction and recognition algorithm processing, to boost the recognition rate of communication for BCI system, and finally to establish a substantial theory and experimental support for BCI application. In this paper, different mental tasks for imaging left-right hands movement from 6 subjects were studied in three different time sections （hint keying at 2s, 1s and 0s after appearance of arrow）. Then we used wavelet analysis and Feed-forward Back-propagation Neural Network（BP-NN） method for processing and analyzing the experimental data of off-line. Delay time △t2,△t1 and △t0 for all subjects in the three different time sections were analyzed. There was significant difference between △t0 and △t2 or △t1 （ P 〈 0.05）, but no significant difference was noted between At2 and At1 （ P 〉 0.05 ）. The average results of recognition rate were 65 % ,86.67 % and 72 %, respectively. There were obviously different features for imaging left-right hands movement about 0.5-- 1s before actual movement ; these features displayed significant difference. We got higher recognition rate of communication under the hint keying at about ls after the appearance of arrow. These showed the feasibility of using the feature signals extracted from the project as the external control signals for BCI system, and demon-strated that the project provided new ideas and methods for feature extraction and elassifieation of mental tasks for BCI.