基于成本-效率优化模型研究人脑结构、功能和代谢的关系

It is very important for understanding the human brain by exploring the relationship among the structural connectivity, functional segregation and integration and metabolic energy consumption. With the combination of advanced brain magnetic resonance imaging (MRI) and complex network approaches, previous studies have made significant progress in characterizing complex properties and their impact on functions and energy consumption. However, it remains poorly understood about the mechanisms underlying the formation of such complex human brain connectome and their relationship with the functions and metabolic energy consumption. In this project, we aim to study how the human brain connectome from multi-modal imaging data is subject to basic constraints. Instead of applying the complex network methods to obtain the network characteristic by phenomenological description, we propose to build the trade-off model using combinatory optimization under two basic constraints – the wiring cost and processing efficiency. By comparing the connections of different brain areas from the model results to that in the real human structural networks, we can identify two special groups of areas, which strongly violate the constraints or obey the constraints. Since the wiring cost constraint is related to the energy consumption, the processing efficiency effects the high reaction of the brain, the two groups of areas may reflect the advanced needs of functions, or be restrained by the metabolic energy consumption. Then we will systematically study the roles played by the two special groups of areas on functional segregation and integration and analyze the relationship between their structural connectivity and metabolic energy consumption by several computational models, so as to obtain the deeper understanding of the structure, function and energy relationship for human brain. The methods and results from this project are expected to help us understand the vulnerability in the human brain derived by the trade-off between the complex functions and the limited energy consumption.

理解脑的结构、功能和代谢三者的关系是脑科学领域的一个重要问题。近十年来，研究者通过将多模态神经影像数据和复杂网络分析方法结合，揭示脑结构和功能网络分别具有一些重要拓扑属性，如模块化组织和核心区域。然而脑结构连接模式与脑功能的分化整合以及能量代谢的关联还知之甚少。本项目拟基于多模态神经影像数据，采用多重约束平衡的优化模型，系统考察脑结构连接与功能、代谢的关系，主要包括：（1）基于多重约束（连接成本和传输效率的平衡）重构人脑结构连接，并识别高度遵守或违反约束条件的特殊脑区；（2）考察特殊脑区对脑功能分化及整合的贡献；（3）探索特殊脑区的代谢率特征，并基于结构连接模式预测脑区能量代谢。本项目将深入揭示脑功能分化整合与能量代谢的关联及其结构基础，为理解大脑工作机理提供全新视角。此外，通过成本效率优化模型识别大脑特殊区域，有望为将来神经退行性疾病的病源机理探索提供重要的方法学支撑。

基于多模态影像数据的脑连接组学和高精度正电子扫描技术积累了丰富的脑结构连接、脑功能静息态数据以及不同脑区代谢水平的数据。然而目前人脑全脑结构、功能和代谢三者之间的关系仍旧不清楚。本项目采用高时空分辨率的多模态影像数据以及脑代谢数据，并开发多种计算神经模型，探索了人脑的结构连接、功能整合和不同脑区重要代谢量水平之间的关系，并揭示了人脑在完成高级整合功能同时避免引入过度脆弱性的收益-风险平衡机制。该项目资助下取得的重要结果包括：.（1）.在猕猴大脑的脑白质纤维网络上采用成本效率生物约束条件的平衡计算模型，发现猕猴大脑近2/3的白质纤维连接服从成本效率约束条件的平衡，并且发现近1/10的脑区连接违反成本效率的生物优化平衡，而这些违反平衡的连接对猴脑的功能分化和整合发挥重要作用。.（2）.基于脑区微观指标不同脑区的神经元细胞的密度、脑区的三维空间位置以及传统的猕猴大脑入侵性低连接密度的结构连接网络，发展多次迭代多因素脑白质纤维预测模型，首次得到猕猴全脑高连接率的白质纤维连接网络，该网络是迄今最完整的灵长类大脑全脑高连接率连接网络。.（3）.基于猕猴大脑入侵性高精度数据发展灵长类脑区纤维连接总长度预测模型，并应用于人脑结构影像数据的结构连接网络，探索其于功能影像数据得到的脑区功能整合能力和脑代谢数据得到的重要代谢指标有氧糖酵解代谢率的关系，发现脑糖酵解代谢水平的结构基础-高代谢脑区有着更长的纤维连接总长度和更高的功能整合能力，并且有着最优化的连接结构的设计来避免引入过度脆弱性，从而进一步揭示人脑的收益-风险平衡的重要机制，推动人脑研究的重要发展。.在项目资助下，共发表（含接收）标注该基金的论文2篇，皆为第一作者或通讯。.作者（含并列）论文2篇（2篇IF>4.4），这包括发表在医学1区杂志Cereb Cortex（接收还未正式发表）以及计算生物学1区杂志PLoS Computational Biology本领域主流国际期刊的研究论文。第三项成果已完成撰写将投国际顶级期刊。