任务态和静息态fMRI数据独立成分分析方法研究

发布时间：2018-05-09 11:35

本文选题：独立成分分析 + 功能核磁共振成像　；参考：《太原理工大学》2014年硕士论文

【摘要】：功能磁共振成像(functional Magnetic Resonance Imaging,fMRI)是上世纪九十年代初发展起来的脑功能成像技术,由于其无放射性、无创伤性、可重复性,与其他非介入性脑功能定位技术相比较具有较高的时间分辨率和空间分辨率,所以很快就成为探测大脑活动最有效的技术。近些年来,功能磁共振成像技术得到了飞速的发展和广泛的应用,尤其在脑科学、神经科学和认知科学中,已经从研究人类大脑的运动、感觉、视觉和听觉等功能,扩展到研究人类的认知、情感、思维、智能、学习等功能。此外,功能磁共振成像主要还应用于临床医学诊断等。功能核磁共振成像数据的主要处理方法有模型驱动方法和数据驱动方法,独立成分分析方法是常用的数据驱动方法。本文研究任务态和静息态下功能磁共振成像的独立成分分析方法。首先将独立成分分析方法应用于颜色和形状的特征捆绑中,得出参与捆绑认知的各大脑区域的主要作用,为建立视觉特征捆绑的认知模型提供理论基础。其次将独立成分分析算法应用于抑郁症的诊断和识别,分析抑郁症患者和正常人的脑区之间的异常,为抑郁症提供辅助诊断方法。主要工作如下： 1.针对研究目的,分析和比较常用的独立成分提取算法,根据采集的fMRI数据选择出最适合的算法,并对独立成分的选取方法进行了说明。 2.研究任务态fMRI数据独立成分分析方法,利用独立成分分析方法研究颜色和形状的特征捆绑问题。采集任务态fMRI数据,对预处理后的fMRI数据提取独立成分,使用多元回归分析方法和相关分析方法选取任务对应的独立成分,使用认知减法进行独立成分之间的计算,得出进行特征捆绑的主要大脑活动区域,进一步分析得出进行颜色和形状特征捆绑任务时大脑区域的激活情况和认知过程,为揭示人的视觉感知过程奠定理论基础,为建立视觉特征捆绑的认知模型提供理论支撑。 3.研究静息态fMRI数据独立成分分析方法,利用独立成分分析方法研究抑郁症患者和正常人的fMRI数据。采集静息态下抑郁症和正常人fMRI数据,将数据分为训练组和测试组、训练组又分为抑郁组和正常组,提取小组独立成分,使用Greicius提出的默认网络,分别选择出抑郁组和正常组的参考成分,将测试组每个被试的参考成分和抑郁组、正常组的参考成分进行相关分析,实现正常人和抑郁症的分类,并检测抑郁症与正常人脑功能网络连接有显著性差异的脑区,为抑郁症提供辅助诊断方法。 4.任务态和静息态fMRI数据独立成分分析方法比较。针对任务态和静息态fMRI数据独立成分分析方法,从实验试用范围、实验的设计、独立成分的提取和选择、独立成分分析的目的和结果进行比较。比较得出相关分析和回归分析能有效提取任务态激活的脑区,通过对默认网络的分析提取出静息态下神经或精神类疾病患者发生病变的脑区,为不同状态下fMRI数据的独立成分分析方法选择提供了可参考的依据。
[Abstract]:In recent years , functional magnetic resonance imaging has been widely used in the research of human cognition , emotion , thinking , intelligence , learning and so on .

In this paper , the independent component analysis method of functional magnetic resonance imaging in task state and resting state is studied . First , the independent component analysis method is applied to the characteristic binding of color and shape , and the main function of each brain region participating in binding recognition is obtained . The theory basis is provided for the cognitive model of the binding of visual characteristics . Secondly , the independent component analysis algorithm is applied to the diagnosis and identification of depression , which can provide auxiliary diagnosis method for depression . The main work is as follows :

1 . For the purpose of research , we analyze and compare the commonly used independent component extraction algorithm , choose the most suitable algorithm according to the data collected , and make a description of the selection method of the independent component .

2 . To study the method of independent component analysis of the data of task state , and study the characteristic binding problem of color and shape by using independent component analysis method . The independent component of the task is selected by using multiple regression analysis method and correlation analysis method . The main brain activity region of the characteristic binding is obtained by using multiple regression analysis method and correlation analysis method .

3 . The independent component analysis method was used to study the data between depression group and normal group . The data were divided into training group and test group . The data were divided into two groups : depression group and normal group .

4 . The independent component analysis method is compared between the task state and the data independent component analysis . The results are compared between the experimental trial range , the design of the experiment , the extraction and selection of the independent component , the purpose and the result of the independent component analysis . It is concluded that the correlation analysis and regression analysis can effectively extract the brain regions activated by the task state . By analyzing the analysis of the default network , the brain region of the lesion of the nerve or the mental diseases in the resting state can be extracted , and the reference basis is provided for the independent component analysis method selection of the MRI data in different states .

【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R445.2;TN911.7

【参考文献】