Extracting Brain Disease-Related Connectome Subgraphs by Adaptive Dense Subgraph Discovery

Abstract

Group-level brain connectome analysis has attracted increasing interest in neuropsychiatric research with the goal of identifying connectomic subnetworks (subgraphs) that are systematically associated with brain disorders. However, extracting disease-related subnetworks from the whole brain connectome has been challenging, because no prior knowledge is available regarding the sizes and locations of the subnetworks. In addition, neuroimaging data is often mixed with substantial noise that can further obscure informative subnetwork detection. We propose a likelihood-based adaptive dense subgraph discovery (ADSD) model to extract disease-related subgraphs from the group-level whole brain connectome data. Our method is robust to both false positive and false negative errors of edge-wise inference and thus can lead to a more accurate discovery of latent disease-related connectomic subnetworks. We develop computationally efficient algorithms to implement the novel ADSD objective function and derive theoretical results to guarantee the convergence properties. We apply the proposed approach to a brain fMRI study for schizophrenia research and identify well-organized and biologically meaningful subnetworks that exhibit schizophrenia-related salience network centered connectivity abnormality. Analysis of synthetic data also demonstrates the superior performance of the ADSD method for latent subnetwork detection in comparison with existing methods in various settings.

A new Mendelian Randomization method to estimate causal effects of multivariable brain imaging exposures

“…(1) Find a submatrix W ⊂ W by a greedy search algorithm 23 to approximately maximize the objective function. (2) Subtract the average of W from each of its entries in W .…”

“…Our method primarily selects a set of exposures that share a common set of IVs guided by data-driven submatrix identification algorithms. 23,24 This method integrates the most informative features from exposures while reducing the burden of horizontal pleiotropy introduced by including too many exposures and IVs simultaneously in the MR model. In this study, we illustrated the application of our method using data from the UK Biobank (UKB) to examine the causal effects of white matter microstructure integrity (WM) measured with factional anisotropy (FA) on cognitive function.…”

Efficient optimization of network metrics in large uncertain graphs

“…brain regions responding to stimuli [130] or related to diseases [205], and commercial value motifs in financial domains [59]. They also have wide applications in graph compression and visualization [30,213,214], indexing for reachability and distance queries [42,107], and social piggybacking [84].…”

“…Dense subgraphs in brain networks can represent brain regions responding together to stimuli [130] or related to diseases [205].…”

“…The discovery of dense subgraphs has attracted extensive attention in the data management community [87,35,196,86,66]. They may correspond to communities [58], filter bubbles and echo chambers [8,126] in social networks, brain regions responding to stimuli [130] or related to diseases [205],…”