脑电信号非线性处理方法在精神分裂症诊断中的应用

发布时间：2018-06-07 04:45

本文选题：精神分裂症 + 脑电　；参考：《兰州大学》2012年硕士论文

【摘要】：精神分裂症是一种严重影响人类健康的精神疾病,随着科技发展,经过不断的探索和改进,如今以脑电生理学技术为基础,以信号处理方法为工具的脑电信号辅助诊断方法已经得到非常广泛的普及。在研究中人们发现脑电信号具有混沌属性,线性的信号处理方法如时频分析、相关分析等无法反映脑电信号的这一性质。而非线性方法以非线性动力学理论为基础,更能够反映脑电信号的本质特性。我们将非线性理论应用到脑电信号处理之中,寻找有效的反映精神分裂症脑电信号本质并且能将其与正常人脑电信号相互区分的非线性特征,从而可以建立一个相对客观的基于脑电信号分析的精神分裂症辅助诊断方法。本文研究并详细介绍了关联维数、最大李雅普诺夫指数、LZ复杂度、CO复杂度、柯尔莫哥洛夫熵等具有代表性的非线性特征算法,并且针对各个特征算法的特性和优缺点进行了比较。根据现有文献研究进度,C0复杂度尚未用于精神分裂症脑电信号的研究之中,本研究中为首次使用。其次我们还选择了关联维数、柯尔莫哥洛夫熵、LZ复杂度与C0复杂度共同应用于比较精神分裂症患者与正常人的脑电信号的区别,对各个特征进行比较并且相互验证结果。我们使用统计方法对计算得到的特征值进行分析,每个特征均有良好效果,除了少部分导联,两组人群的特征值结果在大多数导联上均有显著性差异,并且通过均值比较,病人的非线性特征值要比正常人的高。这个结果说明我们选取的特征能够满足区分两类人群的要求,并且发现C0复杂度具有更良好的结果以及更快的运算速度。同时也验证了相关文献中精神分裂症患者的脑电信号具有更强的非线性和复杂性的结论。我们对这个结果进行了可信度讨论。最后通过分类器得出了91%的分类率,这个结果说明我们所用的基于脑电信号非线性分析的方法适用于鉴别精神分裂症。
[Abstract]:Schizophrenia is a mental disease that seriously affects human health. With the development of science and technology, and through continuous exploration and improvement, it is now based on electrophysiologic techniques. The method of EEG aided diagnosis based on signal processing has been widely used. In the study, it is found that EEG signals have chaotic properties, and linear signal processing methods such as time-frequency analysis and correlation analysis can not reflect this property of EEG signals. The nonlinear method is based on the theory of nonlinear dynamics and can reflect the essential characteristics of EEG. We apply nonlinear theory to EEG processing to find nonlinear characteristics that reflect the nature of EEG in schizophrenia and distinguish it from normal EEG. Therefore, a relatively objective method for the diagnosis of schizophrenia based on EEG analysis can be established. In this paper, some typical nonlinear feature algorithms, such as correlation dimension, maximum Lyapunov exponent LZ complexity and CO complexity, Kolmogorov entropy and so on, are studied and introduced in detail. At the same time, the characteristics, advantages and disadvantages of each feature algorithm are compared. According to the current literature, the complexity of C0 has not been used in the study of EEG in schizophrenia, and it is the first time in this study. Secondly, we choose correlation dimension, Kolmogorov entropy LZ complexity and C0 complexity to compare the differences of EEG between schizophrenic patients and normal subjects. We use the statistical method to analyze the calculated eigenvalues, and each feature has a good effect. Except for a few leads, the results of the eigenvalues of the two groups are significantly different in most leads, and the mean values are compared. The patient's nonlinear characteristic value is higher than that of the normal person. The results show that the selected features can meet the requirements of distinguishing the two groups of population, and it is found that the C0 complexity has better results and faster operation speed. It also verifies the conclusion that the EEG of schizophrenic patients has stronger nonlinearity and complexity. We discussed the credibility of this result. Finally, the classification rate of 91% is obtained by the classifier, which shows that the method based on the nonlinear analysis of EEG signals is suitable for the identification of schizophrenia.
【学位授予单位】：兰州大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TN911.7;R749.3

【参考文献】