基于排序熵的阿尔兹海默症特征提取与分析

发布时间：2018-03-25 13:19

本文选题：阿尔兹海默症　切入点：脑电信号分析　出处：《天津大学》2016年硕士论文

【摘要】：阿尔兹海默症(Alzheimer’s disease,AD)是一种神经系统退行性疾病,起病隐匿,进程缓慢但不可逆,其临床诊断非常困难,利用脑电图(EEG)对AD进行早期诊断和评估已成为当前的研究热点。本文利用加权排序熵(WPE)和多变量多尺度加权排序熵(MMSWPE)来量化AD的脑电信号复杂度,并实现AD脑电分析与评估系统。为了更准确地量化AD病人有噪声干扰的脑电信号复杂度,利用WPE分析了14例重度AD病人及14例同龄健康正常人的16导脑电数据。通过神经元集群模型产生的仿真数据和脑电数据,证实了WPE相较排序熵(PE)的优势。模型仿真结果表明WPE对噪声更敏感。脑电分析结果表明:尽管与正常人相比,AD病人的脑电信号的平均PE值和平均WPE值在各个脑电子频带均有下降,特别是在θ频带,但是WPE能够在各个频带发现更多复杂度显著性降低的导联。为了更准确地量化AD病人多导的脑电信号复杂度,提出了MMSWPE算法。通过洛伦兹系统模型及有色噪声验证了该方法的有效性。脑电数据分析结果表明:与正常人相比,病人在θ和α频带的颞叶区、顶枕区,以及θ、α和β频带的右前额到左顶枕区域的复杂度显著性降低,上述结果反映了AD病人的脑功能失常。AD的脑电分析与评估系统基于MATLAB GUI平台设计并实现,能够实现脑电数据的导入、图像显示、数据分析和文件存储等功能。其中,数据分析包含滤波等预处理功能,PE等复杂度分析功能,及可视图等复杂网络功能。该系统的实现为解决国内缺乏针对于辅助评估AD等退行性精神疾病的软件平台的问题提供途径,为医学工作者提供了一种直观、便捷、可量化的辅助分析手段。
[Abstract]:Alzheimer's disease (Alzheimer 's disease, AD) is a neurodegenerative disease, insidious onset, slow but irreversible, the clinical diagnosis is very difficult, using electroencephalogram (EEG) for early diagnosis and evaluation of AD has become a research hotspot. This paper uses entropy weighted ranking (WPE) and multi scale variable weighted sorting entropy (MMSWPE) EEG signals to quantify the complexity of AD and realize AD EEG analysis and evaluation system. In order to accurately quantify AD patients with EEG signal noise complexity was analyzed by WPE. 16 channel EEG data of 14 cases of severe AD patients and 14 cases of healthy normal people. Produced by neurons cluster model simulation data and EEG data confirmed that WPE (PE) compared with the sort entropy advantage. The simulation results show that WPE model is more sensitive to noise. EEG analysis shows that compared with normal subjects, AD patients The average PE of the EEG signal value and the average WPE value decreased in all brain electronic band were especially in the theta band, but WPE can find more complexity significantly reduced the lead in each frequency band of EEG signals. In order to more accurately quantify AD patients by the complexity of MMSWPE algorithm is proposed by. Lorenz colored noise system model and verify the effectiveness of the method. The EEG data analysis results show that: compared with normal people, patients in the theta and alpha frequency band of temporal lobe, occipital region, and theta, alpha and beta bands right forehead to the left parietal occipital region significantly reduced the complexity. The results reflect the brain function disorder of.AD patients with AD EEG analysis and evaluation system design and Implementation Based on MATLAB GUI platform, to achieve the introduction of the EEG data analysis and image display, data file storage and other functions. The data analysis included filtering etc. The pretreatment function, complexity analysis function of PE, and the visual complex network function. The realization of the system in order to solve the lack of domestic needle method for auxiliary evaluation AD degenerative mental illness of the software platform for medical workers, provides an intuitive, convenient, auxiliary quantitative analysis methods.

【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：R749.16;TP391.41

【相似文献】