CASIA OpenIR  > 毕业生  > 硕士学位论文
基于多中心磁共振影像的阿尔茨海默病脑网络异常表征研究
曲怡达
Subtype硕士
Thesis Advisor左年明
2022-05-14
Degree Grantor中国科学院自动化研究所
Place of Conferral中国科学院自动化研究所
Degree Name工学硕士
Degree Discipline模式识别与智能系统
Keyword阿尔茨海默病 弥散张量成像 白质纤维束 结构连接网络 结构功能耦合
Abstract

阿尔茨海默病(Alzheimer's disease, AD)是一种表现为记忆、情感、认知和行为失常的神经退行性疾病。当前全世界AD患者量大、就诊率不足,且AD尚无有效的治疗措施,因此早期识别和早期干预对于延缓病情至关重要。大脑以网络的组织形式进行信息处理,支撑高级认知活动,既往研究揭示了AD是一种失连接综合征,其大脑发生了网络的而非局部的损伤,因此,刻画AD脑网络异常模式对寻找其客观定量的影像标记物以实现 AD 的早期识别和干预尤为关键。

既往研究大多基于单中心小样本单模态数据集,无法鲁棒评估AD的脑网络异常模式,忽视了AD中结构网络与功能网络耦合关系的变化。因此,本论文旨在使用多站点大样本多模态的磁共振成像数据(865例被试,321例AD患者,265例轻度认知障碍患者,279例正常老人),系统评估AD的白质纤维束异常模式、结构连接网络异常模式、结构功能网络耦合关系的异常改变,分析AD脑网络改变与个体认知能力的关系,并利用机器学习方法评估脑结构连接特征在AD辅助识别中的可行性。论文的主要工作归纳如下:

1. AD 纤维束异常模式研究

白质病变是AD大脑改变的重要表现,纤维束异常是造成脑结构连接网络异常的重要因素,由于缺乏大样本多站点数据集,以往AD的白质纤维束异常模式结果并不一致。因此,为系统评估AD白质纤维束异常的稳定性和泛化性,本文在多站点数据集上,使用自动纤维束量化方法量化了大脑中 20 条白质纤维束的完整性,采用荟萃分析方法识别出AD纤维束的逐点异常模式,发现AD患者存在广泛的纤维束异常改变,纤维束完整性与个体认知能力显著相关,扣带束、胼胝体等是 AD 中存在重要改变的纤维束,这些纤维束的异常会造成大脑内默认网络、视觉网络等功能网络的连接异常。独立站点交叉验证的结果显示异常纤维束特征在AD个体化预测中能达到76.5%的站点平均精度。

2. AD 的结构连接网络异常模式及结构-功能耦合关系异常改变

AD是一种失连接综合征,大量研究表明AD存在脑网络异常改变,但受到样本量、网络构建方法差异的影响,AD的结构连接异常模式尚不清晰,且缺乏结构异常模式和功能异常模式的耦合关系研究。因此,本工作首先从网络连接强度和网络拓扑结构两个方面评估了AD结构连接网络的异常,对比了结构和功能两个模态的脑网络异常模式的异同与关联。研究发现AD结构连接网络的拓扑整合显著降低,对应于默认网络、额顶网络、视觉网络等功能网络的连接强度显著降低,皮下核团内部(包括丘脑、基底神经节等)的连接强度显著增强,这一改变模式在结构网络和功能网络中具有一致性,且与个体认知能力显著相关。AD中结构-功能耦合关系异常增强,提示了AD的脑功能灵活性的降低。最后,独立站点交叉验证结果显示异常结构连接特征在AD的个体化预测能达到75.6%的站点平均精度。

本文从纤维束异常模式的角度为AD的失连接理论提供了证据,清晰地刻画了AD结构连接网络的异常模式,探索了结构-功能网络异常的耦合关系,评估了结构连接特征作为AD影像标记物的有效性和泛化性,为理解AD异常连接模式及其与认知的关系奠定了基础。

Other Abstract

Alzheimer's disease (AD) is a neurodegenerative disease characterized by memory, emotional, cognitive and behavioral disorders. Presently, it is an acknowledged fact that there are numerous AD patients with low consultation rate and no effective treatment, thus early recognition and intervention are crucial to slow down the disease progress. Human brain process information in the form of network to support high-level cognitive activities. Previous studies indicated that AD is a brain network disconnection syndrome with network-wide damage rather than regional damage. Therefore, characterizing the abnormal patterns of brain network is significant for finding quantitative neuroimaging biomarkers to realize the early recognition and intervention of AD.

Most of the previous studies were based on single-site small-sample single-mode datasets, which failed to robustly evaluate the abnormal pattern of brain networks, and ignored the changes in the relationship between structural and functional network in AD. Thus, the current paper aimed to use a large multi-site multimodal MRI dataset (865 subjects, 321 AD patients, 265 mild cognitive impairment patients, 279 controls) to systematically investigate the abnormal pattern of fiber tracts, structural connection network (SCN) and structure-function coupling in AD, and to analyze the relationship between brain network changes and individual cognitive ability, as well as to evaluate the feasibility of structural connection features in the AD identification using machine learning method. The main contributions of this paper were summarized as follows:

1. The abnormal pattern of white matter fiber tracts

White matter disruption is a critical brain change and also a main factor causing the disconnection of SCN in AD. Due to small sample size and various image processing pipelines, previous researches of white matter abnormality patterns in AD are inconsistent. Therefore, in order to evaluate the stability and generalization of the abnormal pattern of fiber tracts, based on multi-site dataset, automated fiber quantification was applied to quantify the integrity of 20 fiber tracts, meta-analyses were used to identify the point-wise changes along tracts. This study found that fiber tracts impaired widely, especially in the cingulum cingulate and callosum forceps, which possibly caused disconnections in the default-mode network and visual network in AD. The diffusion metrics of the altered regions were significantly correlated with cognitive ability. The prediction of individual diagnosis status using tract-based features achieved an site-averaged accuracy of 76.5% through independent site cross-validation.

2. The abnormal pattern of SCN and disrupted structure-function coupling in AD

AD is characterized as a disconnection syndrome, with numerous evidences of disrupted brain network. However, due to the small dataset and various network construction methods, the degenerated pattern of SCN in AD is still unclear, and the studies on the relationship between the structural and functional abnormal patterns were inadequate. Therefore, the current study firstly investigated the AD-associated SCN abnormalities from both connection strength and topological structure aspects, then evaluated the relationship between SCN and functional connection network (FCN). Results showed that in AD group, the integration of SCN was significantly reduced, connections within default mode network, frontoparietal network and visual network were weakened, while connections within subcortical nuclei (thalamus and basal ganglion) were enhanced. This abnormal pattern was consistent in SCN and FCN, and significantly associated with individual cognitive ability. The enhanced structure-function coupling suggested the reduction of brain functional flexibility in AD. Finally, The prediction of individual diagnosis status using SCN features achieved an site-averaged accuracy of 75.6% through independent site cross-validation.

The current study provided evidence for the disconnection of AD from the perspective of white matter fiber tract degeneration, clearly depicted the abnormal patterns of SCN, explored the coupling relationship between the SCN and FCN, as well as evaluated the effectiveness and generalization of structural connection features as the neuroimaging biomarker of AD, contributing to further understand the disconnection and cognitive degeneration of AD.

Subject Area模式识别
MOST Discipline Catalogue工学::控制科学与工程
Pages102
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/48776
Collection毕业生_硕士学位论文
脑网络组研究
Recommended Citation
GB/T 7714
曲怡达. 基于多中心磁共振影像的阿尔茨海默病脑网络异常表征研究[D]. 中国科学院自动化研究所. 中国科学院自动化研究所,2022.
Files in This Item:
File Name/Size DocType Version Access License
QYD_0611_签字版.pdf(30502KB)学位论文 限制开放CC BY-NC-SA
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[曲怡达]'s Articles
Baidu academic
Similar articles in Baidu academic
[曲怡达]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[曲怡达]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.