【作者】 葛家怡；

【导师】 王明时； 刘海婴；

【作者基本信息】 天津大学 ， 生物医学工程， 2009， 博士

【摘要】 睡眠是人类的一项基本生命活动,良好的睡眠有助于消除疲劳、恢复体力、保护大脑、增强机体免疫力等。但当今,越来越多的人睡眠质量有所下降,甚至很多人长期受到失眠的困扰,睡眠质量变坏不仅危害个人身心健康,也影响到了社会的方方面面。因此,从工程学上探索一种效果好、无依赖性,并且不干扰正常睡眠的失眠治疗方法,并对其进行实验效果客观指标评价有着重要的意义。评价睡眠质量不仅仅在于睡眠时间的长短,而且还包括入睡的快慢、睡眠的深度,以及睡眠结构是否合理。因此,睡眠分期又成为睡眠研究的前提,而基于睡眠脑电信号的睡眠分期是睡眠分期研究的最重要、最准确的方法之一,这就要求从更深层次上分析和了解睡眠脑电信号。本文的主要工作如下:1.分析了睡眠脑电信号的新特征。用Itakura距离表示了不同时期睡眠模式上的区别;用非线性样本熵对睡眠脑电信号复杂度进行了表示,并说明了随着睡眠的加深,脑电活动复杂度降低,但快速眼动期又接近于睡眠一期;用多尺度熵表征了不同时期睡眠脑电信号随尺度变化的规律;通过多重分形去势波动分析方法的广义赫斯特指数得出,睡眠过程是一个分形过程,并且存在多重结构,但在一定尺度上也表现出去相关性。2.提取睡眠脑电信号的时域、频域和非线性共19个特征组成特征矩阵,用3层结构BP神经网络进行睡眠分期,正确率达到了89.1%;用支持向量机多类分类的方法分类,正确率达到了92.9%,且睡眠Ⅲ期和睡眠Ⅳ期、睡眠Ⅰ期和快速眼动期之间的误分大大降低。3.构建了真实头模型,并对低频磁场在大脑的感应磁场分布进行了仿真。4.进行了低频磁场睡眠刺激实验,用客观指标对比分析了刺激效果,结果睡眠潜伏期缩短9分钟,总睡眠时间延长25分钟,深睡期比例延长9.9%,说明了低频磁场对睡眠的诱导作用。并对低频磁场作用机理和理论模型进行了探讨。更多还原

【Abstract】 Sleep is a kind of basic vital activities of human being. Good sleep is helpful to relieve fatigue, restore physical strength, protect cerebrum, and enhance organism immunity. But now, the sleep quality of more and more people are reducing, even more and more people are suffering from the torture of insomnia. Bad sleep harms not only individuals physically and mentally, but also consequently affects society’s aspects. Therefore, it will have a significant meaning to explore an insomnia therapy which has good effect to sleep, does not disturb the normal sleeping, is non-addictive and is easily estimated with objective parameters.The appraisal of sleep quality not only lies on the length of sleep time, moreover also on the speed of falling into sleep and the depth of sleep, as well as whether the sleep structure is reasonable. Therefore, the sleep staging becomes the premise in sleep studies, staging by Electroencephalogram(EEG) signal is one of the most important and accurate methods, therefore it requests analyzing and understanding sleep EEG in a deeper level.The main works of this dissertation are as follows:1. Several new characteristics of sleep EEG were analyzed. The difference in sleep patterns of different stages was expressed with the Itakura distance. The complexity of EEG was expressed by the nonlinear sample entropy which showed that the value of sample entropy decreased with sleep’s deepening, but the value of the stage of rapid eye movement approached to of stage 1, the change regulation of EEG signal with the scale in different stages was expressed by the multi-scale entropy. It showed that, by the multifractal detrended fluctuation analysis, the sleep EEG was a fractal process and multi-structure existed, but it also displayed de-correlation performance in certain degree.2. In total, 19 characteristic parameters in time domain, frequency domain and nonlinearity were extracted to construct character matrix. The sleep staging accuracy reached 89.1% with three layers structure BP neural network and achieved 92.9% with the multi-classification method classification of support vector machines, and the classification error rates between sleep stage 3 and stage 4, stage 1 and REM reduced greatly.3. Real head model was constructed, and the magnetic distribution in cerebrum induced by low frequency magnetic field was simulated.4. Low frequency magnetic field sleep stimulation experiment was done, and the stimulation effects were comparatively analyzed by objective parameters. The result shows that the sleep latency reduced 9 minutes finally, the total sleep time expands 25 minutes, and the proportion of deep sleep lengthens 9.9%. It shows the effect of low frequency magnetic field to sleep induced. And the low-frequency-magnetic-field action mechanism and the theoretical model were discussed.更多还原