NCS-FO: Collaborative Research: Individual variability in human brain connectivity, modeled using multi-scale dynamics under energy constraints

近年来，人们对人类大脑的连通性及其与脑部疾病的关系的兴趣激增。神经成像中的功能连通性分析有三种一般形式：交叉相关、有向连接的权重和图论方法。尤其是图论方法，在成像时提供了对网络特征的有价值的见解。然而，它们无法确定大脑是如何具有这些功能的，也无法为未来网络进化的估计提供信息。神经和精神疾病往往具有退行性或振荡的时间进程，时间跨度长达数十年；因此，网络进化对于理解为什么两个诊断相同的人表现出明显不同的发育起点和预后至关重要。在最基本的层面上，临床神经科学目前缺乏工具来探索施加在突触上的生物限制如何影响功能连接模式。除其他外，这些限制包括：有限数量的线粒体的总能量转换率所产生的有限的能量资源，为了维持内稳态而平衡兴奋性和抑制性神经递质的需要，以及神经修复机制(例如，炎症、基质金属蛋白酶-9)。我们的长期目标是开发这些工具，首先关注突触-血流动力学范围内的能量限制，原因有三个。1)血糖负荷与许多神经系统疾病有关，包括癫痫、脑癌和痴呆症。2)能量利用很容易通过饮食进行实验操作，并通过CO-2监测进行量化，协议允许转换到动物模型或从动物模型转换到多尺度建模。3)最近的发现将神经连接与新陈代谢支出联系起来。在短期内，我们的重点是确定三项关键原则的可行性，为拟议的工作做准备。首先，我们将进行一项先导性的神经成像研究(36次扫描；N=12，在三种情况下)，以确定我们提出的对能源供应和需求的实验性操纵引发了大脑网络的重组。其次，我们的目标是跨越尺度：展示基于代理的点神经元模拟如何结合施加在人类神经成像水平上的网络结构，并作为不断变化的输入(能源供应和需求)的函数而演变。第三，我们建议开发/调整所需的方法来对fMRI数据和模拟的动态网络进行数学表征。这项基础性工作将使我们能够开展未来的研究，将代谢过程作为突触、胶质和线粒体的函数进行建模，并使用这些模拟来预测作为神经能量消耗函数的fMRI结果的个体变异性。

Recent years have witnessed an explosion of interest in human brain connectivity and its relationship to brain-based disease. Functional connectivity analyses in neuroimaging have taken three general forms: cross-correlations, weighting of directed connections, and graph-theoretic approaches. Graph-theoretic measures, in particular, provide valuable insights into network features at the time of imaging. Yet, they cannot identify how the brain came to have those features, nor can they inform estimation of future network evolution. Neurological and psychiatric illnesses tend to have degenerative or oscillatory time-courses that range over decades; thus, network evolution will be critical to understanding why two individuals with the same diagnosis show markedly distinct developmental onsets and prognoses. At the most fundamental level, clinical neuroscience currently lacks the tools for probing how biological constraints imposed upon synapses impact functional connectivity patterns. These constraints include, among others: limited energy resources as per the aggregate energy conversion rate of a finite number of mitochondria, the need to balance excitatory and inhibitory neurotransmitters in order to maintain homeostasis, and neural repair mechanisms (e.g., inflammation, MMP-9). Our long-range goal is to develop these tools, focusing first upon energy constraints across synaptic-hemodynamic scales, for three strategic reasons. 1) Glycemic load is implicated in many neurological diseases, including epilepsy, brain cancer, and dementia. 2) Energy utilization is easy to manipulate experimentally through diet, and to quantify via CO-2 monitoring, with protocols that permit translation to/from animal models for multi-scale modeling. 3) Recent findings link neural connectivity to metabolic expenditure. In the short-term, we focus upon establishing feasibility for three critical principles in preparation for the proposed work. First, we will conduct a pilot neuroimaging study (36 scans; N=12, under three conditions) to establish that our proposed experimental manipulation of energy supply and demand provokes reorganization of brain networks. Second, we aim to bridge scales: to demonstrate how agent-based simulations of point-neurons can incorporate network structure imposed at the level of human neuroimaging, and evolve as a function of changing inputs (energy supply, demand). Third, we propose to develop/adapt methods required to mathematically characterize dynamic networks for both fMRI data and simulations. This fundamental work will position us to conduct future research on modeling of metabolic processes as a function of synapses, glia, and mitochondria, and to use these simulations to predict individual variability of fMRI results as a function of neural energy consumption.