意识障碍患者脑电信号的非线性动力学评价分析

发布时间：2018-06-15 14:01

本文选题：意识障碍 + 脑电信号　；参考：《杭州电子科技大学》2012年硕士论文

【摘要】：意识障碍患者脑电信号的分析和评价是当今康复医学工程研究领域热点课题之一，对意识障碍患者的病理诊断及康复治疗具有重大的意义。脑电信号(Electroencephalogram，EEG)是一种能反映人类思维活动的生物电信号，是通过布置在头皮或颅内的电极记录下来的脑细胞群电活动。大量实验已证明脑电信号是混沌信号，具有显著的非线性动力学特征，因此利用脑电信号的各种非线性动力学特征参数可以表达大脑的多种显著性思维意识活动，并可以在临床上利用这些特征参数实现对各种脑功能障碍症状的分析和评价。本文以非线性动力学为理论基础，通过分析脑电信号研究现状及处理方法，设计了能反映大脑思维意识活动的脑电实验；用多尺度Lempel-Ziv复杂度、排列分划Lempel-Ziv复杂度、C0复杂度、基本尺度熵作为刻画脑电信号特征的非线性动力学特征参数，用实验证明了这四种非线性特征参数反映大脑思维意识活动的有效性；探索意识障碍患者和正常人脑电信号的四种非线性特征参数在各种刺激下的变化，并进行对比分析。本文具体完成的研究工作主要包括： (1)从脑电信号的产生机理出发，阐释了脑电信号的分类及处理方法，并设计了两套实验方案：验证性实验和对比实验。验证性实验方案目的是为了证明脑电信号在不同意识任务下具有不同的特征，且相关非线性动力学特征参数能够体现出这些特征，实验共有安静闭眼、闭眼心算、记忆三种模式；对比实验方案目的是分析正常人和意识障碍患者在相同刺激下的脑电信号的非线性特征值异同以及各自在不同刺激下的变化，共有安静状态、唤名刺激和抬手指令等三种模式。 (2)在脑电信号的预处理方面，针对干扰源产生的噪声，，本文采用基于SURE的小波软阈值方法对脑电信号进行消噪，算法分析表明，基于SURE的小波软阈值方法有比其它阈值方法更好的消噪效果；针对独立源产生的干扰，无法用小波消噪的方法滤除，为此，本文首次将基于最大信噪比的盲源分离算法用于脑电信号的伪迹滤波处理，算法的实验结果表明，分离效果和运行效率均优于常用的FastICA算法和Infomax算法。由之本文得出：基于SURE的小波软阈值方法能较好去除脑电信号中的随机噪声；基于最大信噪比的盲源分离算法能成功分离独立源产生的干扰。 (3)在分析非线性动力学基本理论的基础上，首次将多尺度Lempel-Ziv复杂度、排列分划Lempel-Ziv复杂度、C0复杂度、基本尺度熵等四种非线性动力学特征值运用于脑电思维意识活动的分析，计算了FP1、FP2、P3、P4、F7、F8电极脑电信号在不同思维意识模式下的四种非线性特征值，并进行了对比，用单因素方差分析(One-WayANVOA)，通过统计分析软件SPSS 19.0 for windows实现。上述验证性实验结果表明，这四种非线性特征参数能较好地反映正常人脑电信号在不同思维意识活动模式下的差异。 (4)在临床实验数据处理方面，将多尺度Lempel-Ziv复杂度、排列分划Lempel-Ziv复杂度、C0复杂度、基本尺度熵再次运用于意识障碍患者脑电信号的评价和分析，计算了C3、C4、T3、T4电极脑电信号在不同刺激下的四种非线性特征值，分析了正常人和意识障碍患者在相同刺激下的脑电信号的非线性特征值异同以及各自在不同刺激下的变化，采用单因素方差分析(One-Way ANVOA)和两独立样本t检验分析，通过统计分析软件SPSS 19.0 for windows实现。上述对比实验结果表明，脑电信号运用于意识障碍患者的刺激反应和意识状态分析，这四种非线性动力学特征参数可以成为分析评价的主要依据之一。
[Abstract]:The analysis and evaluation of electroencephalogram (EEG) in patients with consciousness disorder is one of the hot topics in the field of rehabilitation medical engineering. It is of great significance for the pathological diagnosis and rehabilitation treatment of patients with consciousness disorder.
Electroencephalogram (EEG) is a bioelectrical signal that can reflect human thinking activity. It is an electrical activity of brain cells recorded by the electrodes arranged on the scalp or intracranial. A large number of experiments have proved that the EEG signals are chaotic signals and have significant nonlinear dynamic characteristics. The characteristic parameters of linear dynamics can express a variety of significant thinking activities in the brain, and can be used to analyze and evaluate the symptoms of various brain disorders by using these parameters in clinical practice.
In this paper, based on the theoretical basis of nonlinear dynamics, by analyzing the current status and processing methods of EEG signal research, a brain electricity experiment can be designed to reflect the mind consciousness activity of the brain, and the multiscale Lempel-Ziv complexity is used to divide the Lempel-Ziv complexity, C0 complexity and basic scale entropy as the characterization of the EEG. The characteristic parameters show that the four nonlinear characteristic parameters reflect the effectiveness of brain thinking activity, and explore the changes of the four nonlinear characteristic parameters of the patients with consciousness and normal human brain signals under various stimuli.
(1) from the generation mechanism of EEG signals, the classification and processing methods of EEG signals are explained, and two experimental schemes are designed: confirmatory and contrast experiments. The purpose of the confirmatory experiment is to prove that EEG signals have different characteristics under different consciousness tasks, and the related nonlinear dynamic characteristic parameters are capable. With these characteristics, there are three modes of quiet closed eyes, closed eye, heart calculation and memory. The aim of the contrast experiment is to analyze the differences and similarities of the nonlinear characteristic values of the EEG signals of the normal and the conscious patients under the same stimulus, and the changes of their respective stimuli under different stimuli. There are three modes of silence, name stimulation and hand raising. Style.
(2) in the preprocessing of EEG signal, the wavelet soft threshold method based on SURE is used to denoise the EEG. The algorithm analysis shows that the wavelet soft threshold method based on SURE has better denoising effect than other threshold methods, and the noise generated by independent source can not be used in wavelet denoising. In this paper, the blind source separation algorithm based on the maximum signal to noise ratio is used for the first time to filter the EEG signal. The experimental results show that the separation effect and the operating efficiency are better than the common FastICA algorithm and the Infomax algorithm. This paper shows that the wavelet soft threshold method based on SURE can better remove the EEG. The blind source separation algorithm based on the maximum signal-to-noise ratio can successfully separate the interference generated by independent sources.
(3) on the basis of analyzing the basic theory of nonlinear dynamics, four kinds of nonlinear dynamic characteristics, such as multiscale Lempel-Ziv complexity, permutation, Lempel-Ziv complexity, C0 complexity and basic scale entropy, are used for the analysis of brain electrical thinking consciousness, and FP1, FP2, P3, P4, F7, and F8 electrode brain electrical signals are calculated in different thinking consciousness. The four nonlinear eigenvalues under the model are compared with the single factor variance analysis (One-WayANVOA) and the statistical analysis software SPSS 19 for windows. The experimental results show that the four nonlinear characteristic parameters can reflect the difference between the normal human brain electrical signals in different thinking modes of activity.
(4) in the clinical experimental data processing, the multiscale Lempel-Ziv complexity, the arrangement of Lempel-Ziv complexity, the C0 complexity and the basic scale entropy are reused in the evaluation and analysis of the EEG signals of the patients with consciousness obstacles, and the four nonlinear eigenvalues of the C3, C4, T3 and T4 electroencephalogram signals under different stimuli are calculated, and the normal people and the normal people are analyzed. A single factor analysis of variance (One-Way ANVOA) and two independent sample t test were used to analyze the difference of the nonlinear eigenvalues of EEG signals and their changes under the same stimulation. The results of the statistical analysis software SPSS 19 for windows were implemented. These four nonlinear dynamic characteristic parameters can be one of the main bases of analysis and evaluation.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.0;TN911.6

【参考文献】