抑郁症静息态功能脑网络异常拓扑属性分析及分类研究

发布时间：2018-06-25 13:07

本文选题：功能磁共振 + 复杂网络　；参考：《太原理工大学》2013年博士论文

【摘要】：人脑是现实世界中最为复杂的网络系统之一。其复杂性不仅体现在数以亿计的神经元及连接的数量,更体现其在不同尺度下的构成以及这些连接在认知功能、思想、感觉及行为时所表现出来的不同模式。近年来,将复杂网络理论应用在神经认知科学中,利用复杂网络基本原理等方法进行属性分析,以期发现网络基本属性及节点间潜在的拓扑关系。复杂网络理论使我们从一个不同的角度来看待人脑这一复杂系统,也为人脑的研究提供了一个新的方向。本文基于复杂网络理论,探讨静息态功能脑网络构建、分析及比较方法,并在此基础上完成静息态脑网络构建分析软件平台的开发；利用功能脑网络进行网络指标组间比较,从全局属性、局部属性、社团化分析等多角度进行组间差异分析；利用抑郁症作为疾病模型验证上述方法的临床可用性,寻找在脑疾病状态及基因影响下的变化规律,探索发现抑郁症早期诊断的影像学标志,突破脑影像技术在精神疾病临床诊断应用所面临的瓶颈问题；针对网络表征,利用机器学习算法,建立抑郁症辅助模型,以辅助临床诊断应用。本文主要创新工作包括有： (1)提出抑郁症静息态功能脑网络指标差异分析方法,并构建分类模型本文分别对抑郁症患者及健康人群的静息态功能脑网络拓扑属性从多个角度进行刻画及比较分析,寻找组间差异,揭示抑郁症在网络层面的指标变化规律。利用多种机器学习方法,将所发现的差异指标作为分类特征,进行分类模型构建及性能评价。并利用敏感性分析,判定其在分类模型中的贡献度,以验证研究方法的合理性。 (2)利用复杂网络模块划分方法进行静息态功能脑模块划分,并提出抑郁症模块结构差异分析技术本文利用基于贪婪思想的CNM模块划分算法,完成抑郁组及对照组的静息态功能脑网络模块划分,并从模块的组成、模块角色、模块间的连接等多个角度,挖掘抑郁症在模块结构上的差异。最后,利用差异模块指标进行分类研究,以验证方法的可靠性,最高正确率可达到90%以上。 (3)提出基于基因的抑郁症脑网络拓扑属性差异分析技术前人研究证明,基因对于脑网络的拓扑属性则存在不同程度的影响。本文利用功能脑网络方法,挖掘GSK3β基因对于抑郁症患者及正常对照的网络拓扑属性差异,以探讨脑网络的基因基础。 (4)提出抑郁症局部一致性指标差异分类技术,构建分类模型并提出特征评价标准局部一致性方法反映了脑区中某个局部的神经元活动在时间上的一致性和同步性。本文利用局部一致性指标,进行抑郁症组间差异分析。利用机器学习方法,验证局部一致性方法的可靠性,并提出通过敏感性分析方法对所选指标进行量化评价。本文是国家自然科学基金项目《抑郁症fMRI数据分析方法及辅助诊断治疗模型研究》(No.61170136)的主要组成部分。研究工作还得到了山西省教育厅高校科技项目《多模态脑网络拓扑属性分析方法研究》(No.20121003)以及太原理工大学青年基金项目《抑郁症静息态功能脑网络拓扑属性差异分析研究》(No.2012L014)的支持。本文重点研究静息态功能脑网络的构建、分析方法及其软件平台的开发,以及脑疾病状态下脑网络的变化规律,在此基础上探索抑郁症等重大脑疾病早期诊断的影像学标志,并建立辅助诊断模型。这不仅是国际前沿基础科学问题,也是国家重大需求。
[Abstract]:The human brain is one of the most complex network systems in the real world. Its complexity is not only reflected in the number of hundreds of millions of neurons and connections, but also its composition at different scales and the different modes of these connections in cognitive function, thought, feeling and behavior. In recent years, the complex network theory has been applied to the complex network theory. In neurocognitive science, attribute analysis is carried out by means of the basic principles of complex networks, in order to find the basic properties of the network and the potential topological relations between nodes. Complex network theory makes us look at the complex system of human brain from a different angle, and also provides a new direction for the research of human brain.
Based on the complex network theory, this paper discusses the construction, analysis and comparison of resting state functional brain networks. On this basis, the rest state brain network construction analysis software platform is developed. By using depression as a disease model to verify the clinical availability of the above methods, look for the changes in the state of brain disease and the influence of genes, explore the imaging signs of early diagnosis of depression, break through the bottleneck of the application of brain imaging technology in the clinical diagnosis of mental disease, and use the machine for network characterization. The learning algorithm is used to establish the auxiliary model of depression to assist clinical diagnosis.
The main innovative work of this article includes:
(1) put forward the method of differential analysis of resting state functional brain network index, and build a classification model.
This paper depicts and compares the topological properties of the resting functional brain network of the depressive patients and the healthy people respectively, and finds the difference between the groups and reveals the changing rules of the index of the depression at the network level. The classification model is constructed by using a variety of machine learning methods to classify the identified differences as the classification characteristics. The sensitivity analysis is used to determine its contribution in the classification model, so as to verify the rationality of the research method.
(2) use the complex network module partition method to divide the resting state functional brain module, and put forward the analysis technology of depression module structure difference.
In this paper, the CNM module division algorithm based on greedy thought is used to divide the resting state functional brain network module of the depression group and the control group, and the differences in the module structure are excavated from the components of the module, the module role and the connection between the modules. Finally, the difference module index is used to classify and study the methods to verify the method. Reliability, the highest correct rate can reach more than 90%.
(3) put forward the technology of topological difference analysis of brain network based on gene depression.
Previous studies have shown that genes have different effects on the topological properties of brain networks. In this paper, the functional brain network method is used to explore the network topological properties of GSK3 beta gene for patients with depression and normal controls, so as to explore the genetic basis of brain networks.
(4) put forward the difference classification technology of regional coherence index of depression, build classification model and put forward characteristic evaluation standard.
The local conformance method reflects the time consistency and synchronism of a local neuron activity in the brain. In this paper, the local conformance index is used to carry out the difference analysis between the depression groups. The reliability of the local consistency method is verified by the machine learning method, and the quantity of the selected index is measured by the sensitivity analysis method. Evaluation.
This paper is the main component of the National Natural Science Foundation of National Natural Science Project (National Natural Science Foundation), the fMRI data analysis method of depression and the model of auxiliary diagnosis and treatment (No.61170136). The research work has also been obtained by the science and technology project of the University of education of Shanxi Province, the study of the multi-modal brain network topology analysis method (No.20121003) and the Taiyuan University of Technology youth fund. This article focuses on the construction of the resting state functional brain network (No.2012L014), the analysis method and the development of its software platform, as well as the changes of brain network in brain disease state, and explore the early diagnosis of depression and other serious brain diseases on this basis. Imaging markers and establishing auxiliary diagnostic models are not only the international frontier basic science issues, but also the major national needs.
【学位授予单位】：太原理工大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：R749.4;O157.5

【参考文献】