Connecting Brain Networks Across Subjects and Across Modalities

描述(申请人提供)：近年来，对大脑成像数据的网络分析越来越流行。这样的分析使研究人员能够描述大脑的结构或功能组织作为一个整体，而不是只关注信号最强的区域。网络分析允许将网络作为一个整体进行表征，以及每个网络节点如何对网络做出贡献。此外，大脑网络的分层组织可以被描述为紧密互连的节点集群的集合，称为模块。这样的模块通常在空间上与与某些认知和感觉过程相关的大脑区域重合。与其他类型的网络数据相比，脑网络数据具有独特的属性。第一个特性是，可以从多个对象观察到大脑网络的多种实现，这在许多社会或技术网络中是不可能的，因为只有一个感兴趣的网络。其次，大脑网络可以在每个节点(可以是解剖区域或体素)或功能相关的模块上跨受试者对齐。利用这些特性的一种方法是将大脑网络数据结合起来。这使研究人员能够跨受试者调查一致的网络属性。在这项提案中，将开发一种方法框架来一起分析多个网络数据。首先，通过建立多个受试者的群体大脑网络，建立一个群体分析的框架。这可以通过在每个节点(具体目标1)或模块(具体目标2)的受试者之间调整网络来实现。除了合并不同受试者的网络外，还将制定一个框架，以合并不同成像方式的脑网络数据，以检查结构和功能脑网络的异同(具体目标3)。多种结构和功能的脑网络研究表明，在网络结构和关键节点的位置上存在共性。因此，所提出的方法将为研究这种结构-功能关系提供一个非常必要的工具。

DESCRIPTION (provided by applicant): In recent years, network analyses of brain imaging data have increased in popularity. Such analyses allow investigators to describe the structural or functional organization of the brain as a whole, rather than focusing only on the areas with the strongest signal. Network analyses allow characterization of a network as a whole, as well as how each network node contributes to the network. Furthermore, hierarchical organization of brain networks can be described as a collection of tightly interconnected clusters of nodes, known as modules. Such modules often spatially coincide with brain areas relevant to certain cognitive and sensory processes. Compared to other types of network data, brain network data have unique properties. The first property is that multiple realizations of the brain network can be observed from multiple subjects, which is simply impossible in many social or technological networks since there is only one network of interest. Secondly, brain networks can be aligned across subjects at each node (which can be an anatomical area or a voxel) or at functionally relevant modules. One way to take advantage of these properties is to combine brain network data. This enables researchers to investigate consistent network properties across subjects. In this proposal, a methodological framework will be developed to analyze multiple network data together. At first, a framework for a group analysis will be developed by building a group brain network from multiple subjects. This can be accomplished by aligning networks across subjects at each node (Specific Aim 1) or module (Specific Aim 2). In addition to combining networks across subjects, a framework to combine brain network data across image modalities will be developed in order to examine similarities and differences in structural and functional brain networks (Specific Aim 3). Multiple structural and functional brain network studies have indicated commonalities in the network structures and the location of key nodes. Thus, the proposed methods will provide a highly needed tool for investigations of such structure-function relationships.