基于多模态磁共振成像的主观认知下降影像标记物和个体识别研究

Alzheimer's disease (AD) does great harm to public health. Unfortunately, its pathogenesis is unclear and there is no radical treatment for it. So, early diagnosis is the key to delay the progression. AD patients have experienced subjective cognitive decline (SCD) - mild cognitive impairment (MCI) - dementia process. The existing researches mainly focus on the MCI stage. In fact, there has been Aβ deposition and brain gray and white matter changes at the SCD stage, so the diagnosis studies of AD should be brought forward to the SCD stage. However, the research of imaging biomarkers of SCD is scarce, and there are no reports about identifying SCD at the individual level. This problem needs to be addressed urgently. Based on the fact that Aβ deposition could be reflected by the Aβ targeting PET/CT imaging and our prior study of identifying amnestic mild cognitive impairment patients at the individual level, we put a hypothesis that multimodal magnetic resonance data can reflect the gray matter and white matter changes of SCD, and the imaging biomarker of SCD and identifying SCD at the individual level can be solved by analyzing these data using support vector machine (SVM). This project will employ the Aβ-PET/CT to participate in the screening of SCD subjects, then investigate the imaging biomarker by integrating sMRI and DTI information, and use SVM to establish classifier to identify SCD from normal controls at the individual level. On this basis, we will disclose the abnormalities of brain white matter and grey matter of SCD individuals, further reveal the relationship between these abnormalities and the pathological changes and cognitive decline of AD from the perspective of multivariate pattern analysis. This study will provide theoretical basis and reliable screening method of subjects for earlier study of AD.

阿尔茨海默病（AD)危害巨大，机制不清，无根治措施，关键在于早期识别。AD经历了主观认知下降（SCD）-轻度认知障碍（MCI）-痴呆的发展过程。现研究集中在MCI阶段，而SCD阶段已有Aβ沉积等病理变化和脑灰白质改变，AD诊断研究应前移到该阶段。但SCD影像标记物研究匮乏，其个体水平识别尚无报道，亟待解决。基于脑Aβ沉积可通过Aβ靶向PET/CT反映和我们对遗忘型MCI成功个体化识别研究的基础，提出假说：SCD脑灰白质改变可通过多模态磁共振反映，应用支持向量机（SVM）可分析SCD的影像标记物并解决个体识别问题。本项目应用Aβ靶向PET/CT显像参与SCD入组筛选，通过分析SCD与正常对照的结构性磁共振和弥散张量成像数据，应用SVM研究SCD影像标记物，进行个体识别；在此基础上揭示SCD脑灰白质改变及与AD病理生理变化、认知下降的关系，为开展AD早期研究提供理论依据和可靠的受试者筛选方法。

背景：阿尔茨海默病（AD)关键在于早期识别。AD经历了主观认知下降（SCD）-轻度认知障碍（MCI）-痴呆的发展过程，SCD阶段已有Aβ沉积和脑结构及功能改变，AD诊断应前移到SCD阶段。但SCD影像标志物研究匮乏，其个体水平识别亟待解决。内容：应用临床评估、神经心理测评结合Aβ靶向PET/CT或PET/MR筛选出AD临床前期SCD受试者，基于支持向量机（SVM）分析SCD与正常对照（NC）多模态MRI数据，实现个体识别，研究SCD影像标志物，揭示SCD脑结构和功能改变及与AD病理变化、认知下降等关系。结果：招募受试者共914人进行临床评估，根据入选标准，筛选出361人进行神经心理测评，其中符合要求者318人；基于SCDplus标准，得到SCD146人、NC172人，共采集多模态MRI影像数据290人、Aβ靶向PET数据101人。在SCD个体化识别方面：利用SVM基于DTI及fMRI 数据对SCD及NC进行分类准确度达到80.24%；应用SVM基于sMRI多个脑区体积信息作为特征，实现了对SCD与NC的高精度区分，平均分类准确度> 94％，分类精度优于以往研究，初步实现SCD个体化识别。在标志物研究方面：发现SCD脑结构连接拓扑效应紊乱；默认网络（DMN）功能连接模式异常；最具判别能力的脑区主要位于DMN和皮层下结构（SCS）；SCD者的脑部萎缩模式与遗忘型轻度认知障碍（aMCI）类似而与NC不同，进一步支持了SCD是aMCI和正常老化之间的过渡期。在揭示SCD脑改变及与AD病理变化、认知下降等关系方面：提示在AD早期功能变化出现在结构变化之前，随着病理发展，结构连接逐渐显示出比功能连接更强的识别能力；SCD所累及脑区与认知功能高度相关。相关结果发表文章5篇，其中SCI文章2篇。意义：初步解决AD早期SCD影像标志物匮乏问题；揭示SCD脑结构和功能异常及其分布模式与AD解剖、病理、认知下降的关系，为AD早期研究提供理论依据和受试者筛选方法；解决从个体水平客观识别SCD问题，实现高精度区分SCD和NC，以期解决AD早期诊断关口前移问题。