The Human Brain as a Complex System: Investigating the Relationship between Structural and Functional Networks in the Thalamocortical System

大多数大脑功能不是由单个区域完成的，而是由分布在整个大脑中的网络的组合、协调活动完成的。一些神经和精神疾病可能是由这些区域有效沟通能力的崩溃引起的。虽然已经开发了几种不同的方法来理解网络的组成区域或节点如何相互作用，但没有一个综合的框架来结合来自不同技术的信息来给出网络功能的总体图像。如果没有这样一个框架，神经成像技术的进步就不能充分利用解剖和功能连接的特征。这个项目的目的是开发这样一个框架，利用内在的大脑活动，可以定义很好的特征模型网络，从而提供结果的自然验证。大脑网络的节点可以用三种不同的区域间连接定义来识别。结构连通性（SC）描述了区域之间的解剖连接，功能连通性（FC）确定了两个区域的活动是否一致地增加和减少，而有效连通性（EC）试图用脑电图或MRI信号来描述大脑，但产生这些信号的潜在神经元群。这些措施中的每一个都可以使用多种不同的数据采集和分析技术进行估计。例如，SC可以通过扩散张量成像（DTI） MRI扫描来确定，这种扫描对白质束中水的扩散很敏感，或者通过皮质厚度的测量来确定。同样，FC可以通过脑电图（EEG）或功能性磁共振成像（fMRI）测量来计算。了解这些不同的连接测量是如何关联的，以及如何将人类大脑功能和结构的测量结合起来产生一个统一的图像，并不是直截了当的。很少有研究通过多种技术获得这样一项工作所需的高质量数据。一个更复杂的问题是定义模型网络，该模型网络具有足够的复杂性，可以为任何方法开发提供现实的测试，同时具有足够好的特征，可以对开发进行验证。我们将以一种新颖的方式克服这个问题，通过建立几十年的侵入性神经生理学实验，这些实验表征了负责产生丘脑皮质振荡（TCO）的网络，由皮层和丘脑网络节点之间的相互作用产生的电生理事件。TCO可以是正常脑功能的标志（阿尔法节律、睡眠纺锤波和k -复合体），也可以是病理生理学的标志，其中最明显的是全身性癫痫的特征——全身性尖峰波放电。该项目将使用TCO定义的网络来研究大脑连接的不同测量之间的关系，开发和优化新方法，以结合并充分利用可以从非侵入性脑成像数据中提取的所有信息。这些网络的建模和分析将基于图论方法。通过使用这些受限且特征良好的模型网络，我们将能够根据以前的神经生理学数据验证我们的工作，并为神经影像学社区提供通用工具。此外，我们将阐明正常和病理大脑活动的产生，以及这是如何从网络连接模式中产生的。

The majority of brain functions are performed not be single regions but by the combined, coordinated activity of networks distributed throughout the brain. Several neurological and psychiatric disorders may be caused by a breakdown of the ability of these regions to communicate effectively. While several different methods have been developed to understand how the component regions, or nodes, of a network interact, there is no comprehensive framework for combining the information from different techniques to give an overall picture of network function. Without such a framework, advances in neuroimaging techniques which allow the characterisation of anatomical and functional connections cannot be fully exploited. The purpose of this project is to develop such a framework, making use of intrinsic brain activity which can define well characterised model networks, thereby providing a natural validation of the results. The nodes of brain networks can be identified using three different definitions of connectivity between regions. Structural connectivity (SC) describes the anatomical connections between regions, functional connectivity (FC) identifies whether the activity of two regions increases and decreases coherently, while effective connectivity (EC) attempts to describe the brain not in terms of EEG or MRI signals, but the underlying neuronal populations which produce them. Each of these measures can be estimated using multiple different data acquisition and analysis techniques. For example, SC can be determined from diffusion tensor imaging (DTI) MRI scans, which are sensitive to the diffusion of water in white matter tracts, or from measurements of cortical thickness. Similarly, FC can be calculated from electroencephalography (EEG) or functional MRI (fMRI) measurements. Understanding how these different measures of connectivity are related, and how measurements of human brain function and structure can be combined to produce a unified picture, is not straightforward. Few studies have acquired the high quality data with multiple techniques that is required for such an undertaking. A further complication is that of defining model networks which are of sufficient complexity to provide a realistic test of any methodological developments, while being sufficiently well-characterised to allow developments to be validated. We will overcome this issue in a novel way by building on decades of invasive neurophysiological experiments which have characterised the networks responsible for the generation of thalamocortical oscillations (TCO), electrophysiological events that are generated by interactions between cortical and thalamic network nodes. TCO can be hallmarks of normal brain function (alpha rhythm, sleep spindles and K-complexes), or pathophysiology, of which the most obvious are generalised spike-wave discharges, characteristic of generalised epilepsy. This project will use the networks defined by TCO to investigate the relationships between different measures of brain connectivity, developing and optimising new methods to combine and fully exploit all of the information that can be extracted from non-invasive brain imaging data. Modelling and analysis of these networks will be based on graph theoretical approaches. By using these restricted and well-characterised model networks, we will be able to validate our work against previous neurophysiological data, and provide general tools for the neuroimaging community. In addition, we will shed light on the generation of normal and pathological brain activity and how this arises from network connectivity patterns.