龙源期刊网 http://www.qikan.com.cn P300脑电信号的特征提取及分类研究 作者:马也 姜光萍 来源:《山东工业技术》2017年第10期
摘 要:针对P300脑电信号信噪比低,分类困难的特点,本文研究了一种基于独立分量分析和支持向量机相结合的脑电信号处理方法。首先对P300脑电信号进行叠加平均,根据ICA算法的要求,对叠加平均的信号进行去均值及白化处理。然后使用快速定点的FastICA算法提取P300脑电信号的特征向量,最后送入支持向量机进行分类。采用国际BCI 竞赛III中的DataSetII数据进行验证,算法的最高分类正确率达90.12%。本算法原理简单,能有效提取P300脑电信号的特征,对P300脑电信号特征提取及分类的任务提供参考方法。
关键词:P300脑电信号;特征提取;独立分量分析;支持向量机 DOI:10.16640/j.cnki.37-1222/t.2017.10.180 0 引 言 近年来随着世界人口的不断增多和老龄化加剧的现象,肌肉萎缩性侧索硬化症,瘫痪,老年痴呆症等患者的基数也相应增长,给社会及病人家属带来了沉重的负担。而近年来出现的涉及神经科学、认知科学、计算机科学、控制工程、医学等多学科、多领域的脑机接口方式应运而生[1]。脑机接口(brain computer interface,BCI)是建立一种大脑与计算机或其他装置联系的技术,该联系可以不通过通常的大脑输出通路(大脑的外周神经和肌肉组织)[2]。这种人机交互形式可以代替语言和肢体动作,使得恢复和增强人类身体与心理机能、思维意念控制变成为可能。因此在军事目标搜索[3]、飞行模拟器控制[4]、汽车驾驶[5]、新型游戏娱乐[6]以及帮助运动或感觉机能出现问题的残障人士重新恢复信息通信功能[7]等方面均有应用并有巨大潜能。
脑机接口系统的性能主要由脑电信号处理模块决定。脑电信号处理模块的核心由特征提取和分类识别两部分组成。常见的脑电信号特征提取方法很多,针对不同的脑电信号有不同的方法。例如时域分析方法有功率谱分析及快速傅里叶变换(FFT)等,适用于P300、N400等潜伏期与波形恒定,与刺激有严格锁时关系的诱发脑电信号;频域分析方法有自回归模型及数字滤波器等,适用于频率特征明显的运动想象脑电信号;时频域分析方法有小波变换,适用于时频特性随时间不断改变的脑电信号。上述方法实时性较好,使用较为广泛,但不能直接表达EEG各导联之间的关系。空间域特征提取方法有共空间模式法(CSP)、独立分量分析法(ICA)等,该类方法可以利用各导联脑电信号之间的空间分布及相关性信息,一般用于多通道的脑电信号特征提取。 [8-10] 龙源期刊网 http://www.qikan.com.cn 在分类识别方面,BCI中使用的分类方法主要有人工神经网络、支持向量机、K均值聚类、遗传算法等等。支持向量机方法在脑电信号分类中有广泛的应用,在随机性强,非线性的分类识别问题中有较强适应性及较高分类正确率。
1 实验数据介绍 1.1 事件相关电位P300 事件相关电位(Event-Related Potential, ERP)是人们经历某种刺激事件时,大脑在信息加工中所诱发出来的一系列脑电活动在头皮上引起的电位变化,是一种由心理或语言因素参与的特殊的诱发电位,1965年由Sutton首次报道。从头皮记录到ERP有两个特征,首先它的潜伏期与刺激之间有严格的锁时关系,其次它有特定的波形和电位分布。ERP的构成分外源性和内源性两部分,外源性成分包括P100, N100, P200波,潜伏期短,受刺激物理特性的影响较大;内源性成分包括N200, P300波,受心理因素影响较大,和人的注意、记忆等认知过程相关。
其中P300是应用最广泛的内源性事件相关电位,因其潜伏期多在300ms左右,又是正相波,因而得名,故又称P300。目前的研究结果表明,P300是联合皮层活动的结果,与复杂的多层次心理活动(认知过程)有关,是感觉、知觉、记忆、等心理过程的变化反映,是人对客观事物的反应过程。因此,P300是一个不需要靠外部行为判断受试者认知过程的客观指标,也可以说是判断大脑高级功能的一个客观指标。
1.2 实验数据选取 实验数据取自国际BCI 竞赛III中的DataSetII数据。This dataset represents a complete record of P300 evoked potentials recorded with BCI2000 using a paradigm described by Donchin et al., 2000, and originally by Farwell and Donchin,1988[11]. In these experiments, a user focused on one out of 36 different characters. The objective in this contest is to predict the correct character in each of the provided character selection epochs.
The user was presented with a 6 by 6 matrix of characters (see Figure 1). The user’s task was to focus attention on characters in a word that was prescribed by the investigator (i.e., one character at a time) All rows and columns of this matrix were successively and randomly intensified at a rate of 5.7Hz. Two out of 12 intensifications of rows or columns contained the desired character (i.e., one particular row and one particular column). The responses evoked by these infrequent stimuli (i.e., the 2 out of 12 stimuli that did contain the desired character) are different from those evoked by the stimuli that did not contain the desired character and they are similar to the P300 responses previously reported. 龙源期刊网 http://www.qikan.com.cn We collected signals (bandpass filtered from 0.1一60Hz and digitized at 240Hz)from two subjects in five sessions each. Each session consisted of a number of runs. In each run, the subject focusedattention on a series of characters. For each character epoch in the run, user display was as follows: the matrix was displayed for a 2.5 s period, and during this time each character had the same intensity (i.e., the matrix was blank).Subsequently, each row and column in the matrix was randomly intensified for 100ms(i.e., resulting in 12 different stimuli一6 rows and 6 columns). After intensification of arow/column, the matrix was blank for 75ms. Row/column intensifications were block randomized in blocks of 12. The sets of 12 intensifications were repeated 15 times for each character epoch (i.e., any specific row/column was intensified 15 times and thus there were 180 total intensifications for each character epoch). Each character epoch was followed by a 2.5 s period, and during this time the matrix was blank. This period informed the user that this character was completed and to focus on the next character in thewordthat was displayed on the top of the screen (the currentcharacter was shown in parentheses).
2 P300脑电信号特征提取 2.1 独立成分分析原理 ICA的基本思想是将多道观察信号按照统计独立的原则通过优化算法分解为若干独立成分,实现信号的增强和分析。设为N个相互独立的信号源,为M个观测信号,其中,。将它们表示为向量的形式,得,。X中的各分量是由S中各独立源经过线图叠而成的,即
上式用矩阵形式可表示为 式中,X为阶常系数混合矩阵;A为未知的系数。要求在A和S未知的前提下,寻找一解混矩阵W,使得
式中,且Y的各分量相互独立,即Y是对源信号S的良好的估计,也可以说S在Y中得到了分离。