【作者】 闫芬；

【导师】 唐一源；

【作者基本信息】 大连理工大学 ， 理论物理， 2006， 硕士

【摘要】 近年来，随着生命科学的发展，特别是以神经信息学为核心的人类脑计划的提出，认知神经科学已经逐渐成为科学研究中的热点，越来越多的学者把计算机科学、理论物理、应用数学的理论和技术用于神经科学的交叉研究中。本文简要介绍了大脑的基本结构和功能以及近年发展起来的功能磁共振成像技术的基本原理及其在脑高级认知功能中的应用，较详细的介绍和应用了基于内部条件的脑区间协方差分析方法和基于推理思维的自适应控制理论分析方法；提出了基于复杂网络理论的介数中心性分析方法。这些方法的研究为有效的处理功能磁共振成像数据提供了方法和现实意义。主要工作如下： 1．为了发现脑区之间相互作用和协同竞争的关系，详细讨论了三种用于功能磁共振数据处理的基于内部条件的脑区间协方差分析(Within-condition Interregional Covariance Analysis，WICA)方法，三种方法分别是基于感兴趣区的有效性连接分析，任务过程中的时间协方差分析和时间依赖性连接的动态分析。WICA方法在获得感兴趣区和脑激活图的基础上，能够进行功能连接分析。本文在分析三种方法的同时，给出了一个语言认知实验的处理结果，结果清晰直观。 2．大脑在执行任务时，推理思维的自适应控制(Adaptive Control of Thought-rational，ACT-R)理论能够预言功能磁共振图像中BOLD响应。本文引入了一个基于ACT-R理论的信息处理模型，对这个模型的适用范围进行了扩大，使它不仅适用于代数方程式的转换而且适用于不等式的比较，同时发现大脑在执行认知任务时体现个体差异性。 3．为了寻找脑内信息流通中的枢纽点和枢纽区域，提出了基于复杂网络理论的介数中心性分析方法。本文基于功能磁共振成像数据建立脑功能网络，通过比较区域介数值和整体平均介数值的关系找到了大脑在执行汉字认知任务时的5个枢纽脑区。同时，发现脑功能网络具有小世界特性，这可以从一定程度上解释大脑信息处理的高速，有效性。 这三种方法都是对脑功能磁共振成像数据处理的研究，但又各有侧重，分别从有效性连接，自适应控制理论，网络拓扑结构角度反应脑机制。更多还原

【Abstract】 Recently, many people use the technology of computer science, theoretic physics and applied math to research neural information in the worldwide. This paper introduces shortly the basal structure and function of the brain and the application of human brain mapping in brain function. These methods of data analysis are presented in detail, which are as follows: within-condition interregional covariance analysis method; an information-processing model based on adaptive control of thought-rational; the betweenness centrality analysis method based on the properties of topological structures of networks. The research on these methods offers theoretic foundation for processing data of functional magnetic resonance imaging. The main work is summarized as follows:1. In order to find connectivity between regions, we present a within-condition interregional covariance analysis method, including ROI-based analysis of effective connectivity, seeding procedure for analysis of temporal covariance and dynamic analysis of the time-dependence of connectivity.2. Predicting the blood oxygenation level-dependent (BOLD) response of functional MRI, researchers have developed an information-processing model based on adaptive control of thought-rational (ACT-R). This paper introduces an ACT-R model and demonstrates it. The result is the ACT-R information-processing model not only meets transformations on equations in an artificial algebra system, but also adapts well to comparisons of the algebra inequality.3. To find the hinge brain areas during the process of information transmission, we construct brain functional networks based on the functional MRI of Chinese character processing and analyze their properties of topological structures. We find five brain areas, which are regarded as hinge brain areas in the task, through compare the relationship between the area betweenness and the average betweenness in BFN. In addition, the BFNs show the small-world effect, which may shed some light on the great efficiency and high speed of brain information processing.These three methods are used for the data processing of functional MRI, but each of them has its advantages. They are based on covariance analysis, adaptive control of thought-rational and the properties of topological structures respectively.更多还原