加权人脑结构网络的模块化算法研究

发布时间：2018-08-13 18:49

【摘要】：人脑是自然界最复杂的系统之一,科研工作者一直致力于使用各种新技术来研究和探索人脑的工作原理和运行机制。近年来基于核磁共振成像的人脑结构网络重构技术日益成熟,利用图论和复杂网络理论去分析人脑网络也正在成为脑科学研究的重点。人脑结构网络是一个复杂网络,存在着模块结构,对大脑整体运行起着至关重要的作用,目前多数研究集中在二值人脑结构网络的模块划分方法上。二值人脑结构网络往往只体现脑区间有无连接的关系,而基于人脑生理信息所构建的加权人脑结构网络则可以表达脑区之间更具体的关系,在此基础上进行的模块结构划分也更有意义。本文主要以加权人脑结构网络为对象,对加权网络的模块化算法展开研究。首先完成了基于核磁共振成像数据的二值人脑结构网络和加权人脑结构网络的构建,并采用Fast Newman算法对二值人脑结构网络进行了模块划分和结果分析;然后,在此基础上对加权人脑结构网络的模块结构划分算法展开了研究,提出了一种基于凝聚节点思想的加权Fast Newman模块化算法,该算法以单个脑区权重和网络总权重为依据构建加权模块度评价指标,并将其增量作为度量值来决定脑区是否合并从而实现模块划分。该算法分别与二值人脑结构网络的模块化算法和现有加权网络的模块化算法进行了实验对比,结果显示,本文算法得到的模块度更高,其模块结构也更贴近已知的人脑生理特征。最后将本文算法应用于精神分裂症患者与健康人的实验数据中,对比实验显示了两组被试的人脑结构网络在模块结构上存在着差异性。
[Abstract]:Human brain is one of the most complex systems in nature. Researchers have been using various new technologies to study and explore the working principle and mechanism of human brain. In recent years, the technology of network reconstruction of human brain structure based on nuclear magnetic resonance imaging (NMR) is becoming more and more mature, and it is becoming the focus of brain science to analyze human brain network by using graph theory and complex network theory. The human brain structural network is a complex network with a modular structure, which plays an important role in the whole operation of the brain. At present, most of the researches focus on the modular partition method of the binary human brain structure network. The binary human brain structure network usually only reflects the relationship between brain regions, and the weighted human brain structure network based on the human brain physiological information can express the more specific relationship between the brain regions. On this basis, the module structure partition is more meaningful. In this paper, the modularization algorithm of weighted human brain network is studied. Firstly, the binary human brain structure network and the weighted human brain structure network are constructed based on the MRI data, and the binary human brain structure network is partitioned by Fast Newman algorithm and the results are analyzed. On this basis, the modular structure partition algorithm of weighted human brain network is studied, and a weighted Fast Newman modularization algorithm based on the idea of condensed nodes is proposed. Based on the weight of a single brain region and the total weight of the network, the algorithm constructs a weighted modular degree evaluation index, and takes its increment as a measure to determine whether the brain region is merged or not, so as to realize the module division. The algorithm is compared with the modular algorithm of the binary human brain network and the existing modular algorithm of the weighted network. The results show that the modular degree of this algorithm is higher than that of the traditional algorithm. Its module structure is also closer to the known physiological characteristics of the human brain. Finally, the algorithm is applied to the experimental data of schizophrenic patients and healthy people. The comparative experiments show that there are differences in the structure of the human brain network between the two groups.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318;O157.5

【参考文献】