基于标准时频变换的脑电信号分析方法  

Analysis method of EEG signal based on normal time-frequency transform

作　　者：周淑伊 刘小燕[1] 周建松[2] 孙刚 Zhou Shuyi;Liu Xiaoyan;Zhou Jiansong;Sun Gang(College of Electrical and Information Engineering,Hunan University,Changsha 410082,China;Mental Health Institute of the Second Xiangya Hospital,Central South University,Changsha 410011,China)

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]中南大学湘雅二医院精神卫生研究所,长沙410011

出　　处：《电子测量与仪器学报》2018年第12期127-133,共7页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(81571341);教育部博士点基金(20130161110010);湖南省研究生科研创新项目(CX2016B128)资助

摘　　要：脑电(EEG)信号是一种非平稳的随机信号,在时间上有很高的分辨率。然而EEG复杂多变并且含有眼电、肌电等多种干扰成分,不利于分析和处理,因此应用一种新的时频变换方法——标准时频变换对EEG信号进行分析。首先采用标准时频变换将EEG信号转换成时频图;然后依照EEG的波段把时频图分成5个部分,将每个部分的总能量值提取出来作为分类特征;再将特征输入K最邻近分类器进行分类;最后以纽约州立大学健康中心提供的酗酒者及正常人的EEG公开数据库为例,来验证方法的有效性。在数据库中选取75名酗酒者和45名正常人进行分类实验,并与其他脑电分类方法进行了对比分析。结果表明,该方法具有更高的分类准确率,准确率可达99%,并且对参数不敏感,所需的分类特征种类少,是一种高效的EEG分类方法,为EEG信号分类提供了新思路。Electroencephalogram ( EEG) is a non-stationary random signal with high resolution in time. However,the EEG signals are complex and variable and contain various disturbing components such as ocular electricity and electromyography which bring difficulties in process and analysis of the signals. Therefore,a new time-frequency transform method called Normal Time-Frequency Transform is applied to analyze the EEG signals. At first,NTFT is used to convert the EEG signal into a time-frequency diagram. Then,according to the EEG rhythm,the time-frequency diagram is divided into five parts. The total energy value of each part is extracted as a classification feature. Then,the feature is input into K-nearest neighbor classifier for classification. Finally,the public EEG database provided by State University of New York Health Center is used to test the effectiveness of the proposed classification method. In the database,75 alcoholics and 45 non-alcoholic subjects are selected for classification,and the results are compared with other classification methods in literature. Results show that the proposed method has higher classification accuracy ( up to 99%) and improved processing efficiency due to less types of features for classification. Moreover,it is not sensitive to parameters and can simplify process of signal analysis. Result of this work provides new thoughts to analyze EEG signal.

关 键 词：脑电信号 标准时频变换 KNN分类器 酗酒识别 

分 类 号：TH77[机械工程—仪器科学与技术]