阿尔兹海默症患者多尺度动态脑网络建模及分析

Alzheimer's disease is an irreversible diffuse brain disease. Early detection and early intervention are of great significance. Brain networks provide a new perspective, however, most of studies only focused on the abnormal topology patterns of patients' brain networks and rarely involved in the dynamic evolution model of brain networks. In this project, the concept of multi-scale dynamic brain network is proposed, which describes the brain network accurately at different time scales from millisecond to progression. This project explores the construction methods and analysis techniques of multi-scale time-varying dynamic brain network to construct multi-layer brain network with different modalities, different frequency bands and different time scales. Based on the theories of the multi-layer network and complex network dynamics, analysis methods of brain networks are improved and developed. And the structural and dynamic characteristics of patients' brain networks are analyzed from topological structure to dynamic evolution.The retest reliability calculation and feature selection algorithm are used to select the stable and significantly different network indicators and excavate the objective imaging indexes. Our purpose is to move the diagnostic window from the mild cognitive impairment to the subjective cognitive decline. Further, according to the evolution model of the brain network from the early stage to the dementia stage to establish the intervention program for the early patients without irreversible structural damage. The purpose of this study is to solve the international frontier basic science problems and this study guides an important research perspective for the national major demand and the Chinese brain program.

阿尔兹海默症是不可逆的弥散类脑疾病，早发现早干预意义重大。脑网络为其提供了新的研究视角，但多数只研究了患者脑网络异常的拓扑模式，较少涉及脑网络的动态演化模式。本项目提出多尺度时变动态脑网络概念，从毫秒到病程进展过程等不同时间尺度精确地描述脑网络。探讨多尺度时变动态脑网络构建方法和分析技术，构建不同模态、不同频段、不同时间尺度的多层脑网络；利用多层网络和复杂网络动力学等理论，完善和发展基于复杂网络理论的脑网络分析方法，从拓扑结构到动态演化多角度分析患者脑网络的结构特征和动力学特征。借助重测信度计算和特征选择算法筛选稳定且差异度较大的网络指标，挖掘客观的影像学指标，将诊断窗口从目前的轻度认知障前移到主观认知下降阶段，进一步依据早期阶段发展到痴呆阶段的脑网络演化模式辅助制定尚未出现不可逆结构损伤的早期患者的干预方案。本研究是国际前沿基础科学问题，也是国家重大需求和中国脑计划的一项重要研究内容。

本项目重点研究了多尺度动态脑网络的构建方法，完善和发展了基于复杂网络理论的脑网络分析框架。通过阿尔兹海默症（Alzheimer’s Disease, AD）全脑动力学建模，探究了AD个体的病变损伤对神经活动、大脑功能以及行为表现的影响，从脑网络、非线性复杂度以及神经动力学方面完善了AD的早期影像学指标体系，为AD早诊提供了基础。.针对传统脑网络建模分析往往忽略了大脑动态变化的认知基础，项目首先构建了AD患者和对照组在不同频段、不同时间窗下的多层网络，对病程的进展过程进行了动态脑网络建模。基于单层脑网络拓扑分析基础，项目完善了多层脑网络的拓扑特征、重要节点、社区划分等分析算法，从多频段、多时间尺度深入挖掘了动态脑网络属性的表征和拓扑结构，并分析了脑疾病状态下的拓扑属性组间特征差异，为挖掘AD神经机制开拓了崭新的视角。为了深入探讨AD病程变化的动力学同步演化机制，为AD早干预提供依据，项目针对大脑的动力学特征，模拟了AD患者局部连接变化对大脑动力学的影响，通过对个体大脑的虚拟损伤，深入研究了局部变化对全脑活动的影响，并发现了个体的损失靶点和病变模式。在AD局部和全局的异常拓扑特征基础上，项目从非线性复杂度、神经动力学以及脑网络等方面扩充完善了多模态影像数据的分析方法，多角度挖掘到AD的异常影像标志物，为进一步分析验证AD患病异常模式和潜在机制提供了指导意义。