醉酒者脑电和正常脑电非线性特性的比较评估  

A Comparative Evaluation of Nonlinear Characteristics of EEG for Alcoholic and Normal People

作　　者：范金锋[1] 邵晨曦[1] 王剑[1] 杨明[2] 王子才[2] 

机构地区：[1]中国科学技术大学计算机科学技术系,合肥230027 [2]哈尔滨工业大学控制与仿真中心,哈尔滨150001

出　　处：《中国生物医学工程学报》2008年第1期18-22,28,共6页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金资助重点项目(60434010);安徽省国际科技合作项目(06088023)

摘　　要：为了刻画醉酒者脑电和正常人脑电之间的多种非线性特征并区分两者之间的差异。采用基于信息理论的排列熵(PE)、奇异值分解(SVD)熵、近似熵(ApEn)和基于混沌理论的相关维(CD)、最大Lyapunov指数(LLE)、Hurst指数,分别对35个醉酒者和35个正常人的脑电时间序列进行计算分析。同时,对选用的6种非线性特征分析方法在这两组数据的分析结果上,运用方差分析(ANOVA)进行统计学的比较评估分析。结果表明,ApEn、CD和LLE这三种方法能够显著区分这两组数据,但由于算法时间复杂度较高,不宜用于实时分析;而PE和Hurst指数方法相对简单、算法复杂度较低、分析结果有效,可以应用于对正确性和实时性要求均较高的EEG数据分析处理中。In order to characterize various nonlinear characteristics of EEG with alcoholic people and the control, some methods are selected for calculation and analysis for comparison. These methods are those based on information theory, such as permutation entropy （ PE）, singular value decomposition （SVD） entropy, approximate entropy （ ApEn）, and other ones based on chaos theory, such as correlation dimension （CD）, the largest Lyapunov exponent （LLE）, Hurst exponent. 35 alcoholic people and 35 control people were included in the experiments respectively. Comparisons based on analysis of variance （ANOVA） were also drawn between these six nonlinear characteristic analysis methods. Results showed that ApEn, CD and LLE were better in distinguishing the EEG between alcoholic people and the control, but they could not be used for on-line analysis due to too much time-consuming. While PE and Hurst methods were with less computmion complexity and effective as well, being available for real-time analysis.

关 键 词：复杂系统 脑电 信息理论 混沌理论 评估 

分 类 号：R319.04[医药卫生—基础医学]