小波变换和小波熵在睡眠脑电信号变化特性研究中的应用价值  

Application of wavelet transform and wavelet entropy in the characteristic study of sleep electroencephalogram signal change

作　　者：肖余粮[1] 和卫星[1] 陈晓平[1] 吉奕[1] 吴秋明[1] 

机构地区：[1]江苏大学电气信息工程学院,江苏省镇江市212013

出　　处：《中国临床康复》2006年第25期118-120,共3页Chinese Journal of Clinical Rehabilitation

摘　　要：目的:利用小波熵对各期睡眠脑电复杂度进行动态时变特性及统计特性的分析比较。方法:实验于2005-08/2005-09在江苏大学电气信息工程学院生物工程系实验室完成。睡眠脑电数据取自MIT-BIH数据库,8名被试者每人取一导整夜脑电信号,无睡眠方面疾病。采用其睡眠数据进行复杂度分析实验研究,小波包去噪信号进行多尺度分解后,利用小波熵求其各睡眠期和4个有用频率带δ、θ、α、β的小波熵值。结果:6种不同状态下脑电的小波熵之间有着明显的区别,其中清醒期小波熵均值最大;在非快速眼动睡眠期的4个分期里S1期小波熵均值最大、S4期小波熵均值最小,随着睡眠的深入呈现出逐渐减少的趋势;快速眼动睡眠期小波熵均值介于清醒期和非快速眼动睡眠期之间。在脑电波4个基本节律带(δ、θ、α、β)下,共同的特点是在β节律下小波熵均值最小,在θ节律下小波熵均值最大;分别在δ、α、β频率带下清醒期小波熵最大,在S1、S2、S3、S4期随着睡眠的深入有逐渐减少的趋势,快速眼动睡眠期介于清醒期和非快速眼动睡眠期之间;在S1、S2、S3、S4期小波熵均值在4个节律带的波动幅度要明显大于清醒期和快速眼动睡眠期。在睡眠各期及各频率带下小波熵方差值变化趋势与均值类似并更明显。结论:小波熵作为一种复杂性测度方法在睡眠各期脑电的应用结果中显示随着睡眠的深入,小波熵逐渐减少,这与理论上是符合的,所以小波熵可以作为不同生理状态下脑电的变化特性的检测指标,既能区分长时间段脑电复杂度之间的差异,又能反应微状态下的变化特性。AIM：To analyze and compare the dynamic features and statistical characteristics of the complexity measure for sleep electroencephalogram （EEG） in each vigilant state with wavelet entropy （WE）. METHODS：This experiment was carried out at the laboratory of bioengineering department of College of Electronic and Information Engineering of Jiangsu University from August 2005 to September 2005. Experimental data were obtained from MIT-BIH database. One channel overnight sleep EEG signal was adopted from one of eight testees. No sleeping illness was found in each testee. Then, we analyzed and studied the complexity of sleep EEG based on experiment sleep data, the wavelet packet de-noising EEG signals were decomposed into four components of 8, 0, α and β by using orthogonal wavelet transform （WT）. Then, the wavelet entropy of four components and the summation were calculated as a function of time. RESULTS： Experimental results showed that there were significant differences among the WEs of EEGs recorded in the six states of sleep stage. There was the maximal WE value in the resting conscious stage. Among the four sleep stage of non-rapid eye movements （NREM）, WE value was the maximal in S1 sleep stage and the minimal in S4 sleep stage, WE value decreased in turn from slight sleep to deep sleep. WE value of rapid eye movement （REM） sleep stage was between resting conscious stage and NREM sleep stage. Among the four basal rhythm bands （δ,θ, α and β） of EEG, the common characteristic was that WE value was minimal in the β band and WE value was maximal in the θ band. In the δ, α and β bands, WE value of resting conscious stage was maximal, WE value decreased in turns from S1 sleep stage to S4 sleep stage, WE value of REM sleep stage was between resting conscious stage and NREM sleep stage. The fluctuated range of WE value of the four basal rhythm bands was obviously larger during S1, S2, S3 and S4 sleep stage than resting conscious stage and NREM sleep stage. Variance value tendency wa

关 键 词：脑电描记术 睡眠 快速眼运动 熵 

分 类 号：R319.3[医药卫生—基础医学]