基于多通道线性描述符的脑—机接口分类算法的研究  

Study on Classification Algorithms of a Brain—Computer Interface Based on Multichannel Linear Descriptors

作　　者：魏庆国[1] 高小榕[2] 王毅军[2] 高上凯[2] 

机构地区：[1]南昌大学电子信息工程系,南昌330029 [2]清华大学生物医学工程系,北京100084

出　　处：《中国生物医学工程学报》2007年第6期810-817,共8页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金资助项目(60318001);北京自然科学基金重点项目(30510001)。

摘　　要：脑电信号的特征提取是脑—机接口(BCI)中的一个关键部分,对提高分类正确率和信息传输率起着决定性的作用。本研究利用多通道线性描述符提取脑电信号的分类特征信息,将三个描述符单独和联合地施加于三个感兴趣的电极子集,导出12个特征矢量。五个受试参加了一个在线反馈BCI实验。实验期间他们被要求想象左手或右手运动,记录的脑电图数据用于离线分析。对来自7导和11导两个电极子集的8个特征矢量,五个受试平均的分类精度在89%和93.5%之间,而最好的分类精度在85%与99.9%之间。比较基于描述符的特征与基于自回归(AR)模型的特征分类性能,结果表明多通道线性描述符是一种有效的特征提取方法。使用该方法提取特征时,理想的电极数应在7与11之间。The feature extraction of electroencephalographic（EEG） signals is a crucial component in a brain-computer interface（BCI）,which plays a decisive role in increasing classification accuracy and consequent information transfer rate.Feature information for classification was extracted by three multichannel linear descriptors that were applied alone and together to three electrode subsets of interest,resulting in twelve feature vectors.Five subjects participated in an on-line feedback BCI experiment during which they were asked to imagine the movement of either left or right hand.The EEG recordings from all subjects were analyzed off-line.The averaged classification accuracies of the five subjects ranged from 89% to 93.5% for the eight feature vectors from the two electrode subsets of seven and eleven leads,and their best classification accuracies ranged between 85% and 99.9%.The performance of the descriptors-based features was compared with that of autoregressive（AR） model-based features.Results indicated that multichannel linear descriptors were effective methods for feature extraction in motor imagery-based BCIs and the ideal electrode number used in the method were between seven and eleven.

关 键 词：脑—机接口(BCI) 脑电图(EEG) 多通道线性描述符 特征提取 支持向量机(SVM) 

分 类 号：R318[医药卫生—生物医学工程]