Joint hub identification for brain networks by multivariate graph inference.
Joint hub identification for brain networks by multivariate graph inference.
复制标题
DOI：
10.1016/j.media.2021.102162
复制
发表时间：
2021-10
影响因子：
10.9
通讯作者：
Wu G
中科院分区：
文献类型：
--
作者：
Yang D;Zhu X;Yan C;Peng Z;Bagonis M;Laurienti PJ;Styner M;Wu G
神经成像的最新发展使我们能够研究活体大脑区域之间的结构和功能连接。越来越多的证据表明，中枢节点在大脑交流和神经整合中发挥着核心作用。然而，这种高度的中心性使得中枢节点特别容易受到病理性网络改变的影响，因此从脑网络中识别中枢节点在神经成像领域引起了极大的关注。当前流行的集线器识别方法通常以单变量的方式工作，即，基于每个节点处的连接性简档的启发式或网络模块的预定义设置来一个接一个地选择集线器节点。由于没有充分利用整个网络(例如网络模块)的拓扑信息，因此当前方法识别链接多个模块的集线器(连接器集线器)的能力有限，并且偏向于识别在同一模块内具有多个连接的集线器(省级集线器)。为了解决这一挑战，我们提出了一种新的多变量中心识别方法。我们的方法将连接器集线器标识为在从网络中移除时将网络划分为断开的组件的集线器。此外，我们扩展了我们的枢纽识别方法，从一组网络数据中寻找基于种群的枢纽节点。我们在模拟数据和人脑网络数据上将我们的中枢识别方法与现有方法进行了比较。我们提出的方法实现了更准确和可复制的中枢节点发现，并在识别与阿尔茨海默病和强迫症等神经疾病相关的网络变化方面显示出更强的统计能力。
Recent developments in neuroimaging allow us to investigate the structural and functional connectivity between brain regions in vivo. Mounting evidence suggests that hub nodes play a central role in brain communication and neural integration. Such high centrality, however, makes hub nodes particularly susceptible to pathological network alterations and the identification of hub nodes from brain networks has attracted much attention in neuroimaging. Current popular hub identification methods often work in a univariate manner, i.e., selecting the hub nodes one after another based on either heuristic of the connectivity profile at each node or predefined settings of network modules. Since the topological information of the entire network (such as network modules) is not fully utilized, current methods have limited power to identify hubs that link multiple modules (connector hubs) and are biased toward identifying hubs having many connections within the same module (provincial hubs). To address this challenge, we propose a novel multivariate hub identification method. Our method identifies connector hubs as those that partition the network into disconnected components when they are removed from the network. Furthermore, we extend our hub identification method to find the population-based hub nodes from a group of network data. We have compared our hub identification method with existing methods on both simulated and human brain network data. Our proposed method achieves more accurate and replicable discovery of hub nodes and exhibits enhanced statistical power in identifying network alterations related to neurological disorders such as Alzheimer’s disease and obsessive-compulsive disorder.
登录
查看更多内容
DOI：
10.1523/jneurosci.1929-08.2008
发表时间：
2008-09-10
期刊：
The Journal of neuroscience : the official journal of the Society for Neuroscience
影响因子：
--
作者：
Bassett DS;Bullmore E;Verchinski BA;Mattay VS;Weinberger DR;Meyer-Lindenberg A
通讯作者：
Meyer-Lindenberg A
影响因子：
5.3
作者：
Achard, S;Salvador, R;Bullmore, ET
通讯作者：
Bullmore, ET
影响因子：
3.7
作者：
Hagmann, Patric;Kurant, Maciej;Gigandet, Xavier;Thiran, Patrick;Wedeen, Van J.;Meuli, Reto;Thiran, Jean-Philippe
通讯作者：
Thiran, Jean-Philippe
影响因子：
3.7
作者：
Halu A;Mondragón RJ;Panzarasa P;Bianconi G
通讯作者：
Bianconi G
影响因子：
3.7
作者：
Alexander-Bloch, Aaron F.;Vertes, Petra E.;Gogtay, Nitin
通讯作者：
Gogtay, Nitin