独立分量分析与小波阈值在癫痫脑电信号去噪中的应用研究

发布时间：2018-03-09 23:25

本文选题：脑电图　切入点：脑电噪声　出处：《西北大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着生物神经科学与信息科学等领域的飞速发展,基于脑电信号的相关研究已经成为重要课题。脑电图是大脑电活动的记录,它包含着大量的生理、心理和病理信息,在研究人脑功能、行为意识、康复工程,尤其在癫痫、阿尔茨海默症等脑部疾病诊断方面的工作有着显著的作用。事实上,从脑电图仪记录的脑电信号极易受到非脑神经组织及周围环境等干扰而形成多种脑电噪声,这些噪声往往会影响脑电信号的处理与分析。因此,关于脑电信号噪声去除方法的研究已成为一个热点领域。尽管目前已有很多研究在去除脑电噪声方面有了很大进展,但许多算法仍存在人工辨别噪声、选取干净参考噪声及辅助一定方法识别脑电噪声等问题。鉴于此,本文提出了基于独立分量分析和小波阈值的脑电信号自动去噪方法,并通过结合已有的特征提取方法和超限学习机完成癫痫性发作的自动检测,从而对所提出的脑电信号去噪方法的可行性与有效性进行了验证。主要研究内容如下:第一章系统阐述了脑电信号去噪方法的研究背景与问题,以及国内外有关脑电信号去噪方法的研究现状。第二章分别介绍了独立分量分析的数学背景、信息论基础及基本模型等相关理论,并对基于独立分量分析的相关算法的原理进行了推导。第三章首先介绍了脑电图的采集过程与脑电信号所具有的基本特征;其次说明了脑电信号噪声去除的过程,主要包括ICA分解、含噪独立源信号的识别、小波阈值方法。第四章阐述了癫痫性发作的自动检测流程,通过数值实验验证和分析了本文提出的脑电信号去噪方法的可行性和有效性。
[Abstract]:With the rapid development of biological neuroscience and information science, the related research based on EEG has become an important subject. EEG is a record of brain electrical activity, which contains a lot of physiological, psychological and pathological information. In the study of human brain function, behavioral awareness, rehabilitation engineering, especially in the diagnosis of epilepsy, Alzheimer's disease and other brain diseases. In fact, EEG signals recorded from electroencephalograph are easy to be disturbed by non-brain nerve tissue and surrounding environment, which often affect the processing and analysis of EEG signals. The research on the methods of EEG noise removal has become a hot field. Although many researches have made great progress in the removal of EEG noise, many algorithms still have artificial noise discrimination. Selecting clean reference noise and auxiliary certain method to identify EEG noise and so on. In view of this, this paper proposes an automatic denoising method of EEG signal based on independent component analysis and wavelet threshold. And through the combination of existing feature extraction methods and out-of-limits learning machine to complete the automatic detection of epileptic seizures, Thus, the feasibility and effectiveness of the proposed method are verified. The main research contents are as follows: in chapter one, the background and problems of the denoising method of EEG signal are systematically described. In the second chapter, the mathematical background, information theory foundation and basic model of independent component analysis (ICA) are introduced. The principle of the algorithm based on independent component analysis (ICA) is deduced. In chapter 3, the process of EEG acquisition and the basic characteristics of EEG are introduced, and the process of EEG noise removal is described. It mainly includes ICA decomposition, recognition of noisy independent source signals, wavelet threshold method. Chapter 4th describes the automatic detection process of epileptic seizures. The feasibility and effectiveness of the proposed method are verified and analyzed by numerical experiments.
【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R742.1;TN911.7

【相似文献】