基于DTI的阿尔兹海默病及轻度认知障碍脑结构网络改变的研究

发布时间：2018-04-17 09:10

本文选题：弥散张量核磁共振图像 + 阿尔兹海默病　；参考：《西安电子科技大学》2015年硕士论文

【摘要】：阿尔兹海默病(Alzheimer’s Disease,AD)被认为是老年人中最常见的痴呆型疾病。AD的发病率很高,但该疾病的初期发展缓慢,往往不易察觉。这种疾病初期表现为记忆力衰退、交流障碍等,后期患者将逐渐丧失活动和记忆能力,直至死亡。AD给患者家属和整个社会带来非常沉重的经济负担。轻度认知障碍(mild cognitive impairment,MCI)表现为一定程度的认知障碍、记忆功能明显出现衰退,常常被认为是介于正常人和痴呆状态之间的一种病态,虽然该病的危害程度不及AD,但随着疾病的发展,会有一定比例的MCI患者转为AD。近几十来,研究人员对这类疾病进行了深入的研究,取得了很多重要的研究成果,但该病的病理机制仍然不能被完全解释,因此需要进一步的研究。随着核磁共振影像技术的发展,通过大脑功能和白质纤维的变化来研究疾病,为研究AD的发生、发展过程提供了新的思路。近年来,有关白质纤维结构改变的研究大多基于感兴趣区域和现有网络模型(如小世界网络和无标度网络)的一些指标的改变上。但由于AD病变的位置遍布全脑,因此基于感兴趣区域的研究并不能很好地反映疾病发展的过程,同时已有的网络结构模型是在现实世界网络中抽象出来的,虽然人脑网络在一定程度上满足这些网络模型的特征,但是,在具体疾病的表征上却不能得到一致性的结论。大量的研究资料表明,AD患者的大脑网络中心节点(hubs)存在明显的失连接现象。因此我们提出假设,用基于hubs构建成的rich-club模型,可以更好的描述AD、MCI病人与正常人脑结构的差异。本文主要探讨了作者在这方面的研究成果。其次,我们利用聚类分析方法对功能核磁数据进一步处理,在提高功能网络性噪比方面进行了有益的探索研究。全文主要的研究结果和创新点可概括为以下三个部分:第一部分:研究AD和MCI在rich-club分级网络中大脑结构的改变。主要基于结构核磁共振图像和弥散张量图像,通过标准模板将人大脑皮层分为90个脑区。经过图像分割、图像配准、纤维束追踪的方法,构建出由90个节点组成的人脑白质纤维连接网络。通过计算该连接网络的rich-club系数,证明三组被试(AD、MCI和正常对照组(NC))满足rich-club网络模型,接着在该等级模型上,统计分析每组被试在不同等级网络连接强度的改变。结果发现AD病人的失连接主要发生在第二级和第三级网络,而MCI的失连接主要发生在第三级网络,而且这两组被试失连接的程度与被试表现出的认知障碍程度呈正相关。这样的结果在某种程度上也说明了AD患者比MCI病情更为严重。第二部分:研究AD的生物学标记。第一部分的研究成果表明,AD患者大脑一些重要节点网络连接发生了明显改变。我们推测通过计算构成rich-club节点的一些特征量,可以找到一些AD的特征性生物标记。我们发现,用左侧壳核的介数中心度、左侧壳核与左侧额下回的连接强度做为网络特征,可以以89.09%的准确度(基于SVM的平均敏感度)将AD病人和正常人区分开。第三部分:利用聚类分析方法在提高功能网络性噪比方面进行探索研究。主要包括了计算模型的基本假设,数据处理以及对模型的验证计算。虽然研究结果不尽如人意,但研究思路具有一定的合理性和启发性,进一步的工作有待进行。
[Abstract]:Alzheimer's disease (Alzheimer 's Disease, AD) is considered to be the most common type of dementia onset.AD disease in the elderly rate is very high, but the early development of the disease is slow, often difficult to detect the disease. The early manifestations of memory loss, communication disorders, late patients will gradually lose activity and memory ability until death,.AD to patients' families and the society brought very heavy economic burden. Mild cognitive impairment (mild cognitive, impairment, MCI) showed certain cognitive impairment, memory function decline, is often considered a morbid state between normal people and dementia, although the disease is less than the degree of harm AD, but with the development of the disease, there will be a certain percentage of MCI were converted to AD. in recent decades, researchers have conducted in-depth research on this kind of disease, has made many important achievements, but The pathological mechanism of this disease still cannot be fully explained, so the need for further research. With the development of nuclear magnetic resonance imaging technology, the brain function and white matter fiber changes to study diseases, to study the occurrence of AD, provides a new way of development. In recent years, the research about model changes of white matter fiber structure most of the existing network and based on region of interest (such as small world networks and scale-free networks) some of the index changes. But due to the AD lesion location throughout the brain, so the study of region of interest can not well reflect the development of the disease process and based on the existing network structure model is in reality world network abstracted, although the human brain network to meet the characteristics of the network model to a certain extent, but cannot get consistent conclusions in the characterization of specific diseases. A large number of research funding The material shows that the brain network center node with AD (hubs) has obvious loss of connection phenomenon. So we propose the hypothesis, using the rich-club model constructed based on hubs, can better describe the differences in AD, MCI patients and normal brain structure. This paper mainly discusses the author in this area of research. Secondly, we using the cluster analysis method further processing of functional magnetic data, improve the function of network in noise ratio was explored. The main research results and innovations can be summarized as the following three parts: the first part: the research of AD and MCI in brain rich-club hierarchical network structure is mainly based on nuclear magnetic resonance structure change. Images and diffusion tensor images, the standard template will be divided into 90 human cortical regions of the brain. After image segmentation, image registration, fiber tracking method, constructed by the 90 node group The human brain white matter fiber connected to the network. By calculating the rich-club coefficient of the network connection, that three groups of subjects (AD, MCI and normal control group (NC)) to meet the rich-club network model, and then the grade model, statistical analysis of each group of subjects in different levels of network connection strength changes. The results show that the connection occurs mainly in the second level and third level network of AD patients, and MCI connectivity occurred mainly in the third level network, and the two groups of subjects lost connection degree and the participants showed the degree of cognitive impairment was positively correlated. This results in a certain extent also shows that AD patients were more severe than the condition of MCI. The second part: Study on biological markers of AD. The first part of the research results show that the AD brains of some important nodes of network connection has changed obviously. We speculate that through the calculation of some characteristics of rich-club node structure The amount, you can find the feature of some biomarkers of AD. We found that with the left putamen betweenness centrality, the connection strength of the left inferior frontal gyrus and left putamen as characteristic of the network, with 89.09% accuracy (based on the average sensitivity of SVM) separated from AD patients and normal people. The third part: by using cluster analysis method in the study of improving ratio function of network noise. Including the basic assumptions of the model, data processing and verification of the model calculations. Although the results are not satisfactory, but the research has certain rationality and enlightening, further work to be done.

【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：R749.16

【相似文献】