离散粒子群优化-贝叶斯线性判别分析算法用于视觉事件相关电位P300的分类  

Classification of Visual Event-related Potential P300 Based on DPSO-BLDA Algorithm

作　　者：张宇[1] 王行愚[1] 张建华[1] 金晶[1] 

机构地区：[1]华东理工大学信息科学与工程学院,上海200237

出　　处：《中国生物医学工程学报》2010年第1期46-52,共7页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金资助项目(60775033;60674089);上海市浦江人才计划项目(07PJ14031);上海市重点学科项目(B504)

摘　　要：P300在头皮上的导联位置并不明确,目前对P300的分类研究中,采用的电极组合各不相同,且不同被试在同一电极组合下得到的分类效果存在一定差异,要使所有分类精度都达到最优比较困难。而采用全导联方式则增加了数据处理量,导致系统实时性要求不能满足。为解决该类问题,提出一种基于离散粒子群优化(DPSO)的算法对P300进行最优电极组合选择,并将其与F-score进行了比较。然后利用贝叶斯线性判别分析(BLDA)对P300进行分类,比较了最优电极组合和其他电极组合下的分类结果,表明了DPSO对脑电最优电极组合选择的有效性,并提出了一组可能普适的P300最优分类电极组合,对提高基于P300的BCI系统实时性有重要意义。Presently, the electrodes configurations used in the process of classifying P300 are various since the electrode sites are uncertain. It is difficult for all the classification accuracies to reach the optimal resuhs because of the difference in the accuracy among different subjects for a certain kind of electrode configuration. The real-time performance of a system declines when the amount of data increases due to the use of whole cerebral method. A discrete particle swarm optimization （DPSO）-based method for the optimal selection of P300 electrodes was proposed and compared with F-score in the paper. The results of optimal electrodes configuration were compared with that of other configurations in a Bayesian linear discriminant analysis （BLDA）-based classification study for P300. This study showed the validity of applying DPSO to the optimal electrodes configuration of EEG potentials. The proposed optimal electrodes configuration for P300 is promising in real-time performance of P300-based BCI.

关 键 词：离散粒子群优化(DPSO) 贝叶斯线性判别分析(BLDA) P300 最优电极选择 分类 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] R318.04[自动化与计算机技术—控制科学与工程]