A Neuropathological Hub Identification for Alzheimer’s Disease Via joint Analysis of Topological Structure and Neuropathological Burden

Abstract

Mounting evidence shows that the neuropathological burden associated with Alzheimer’s disease spreads along the network pathway and is often selectively accumulated at certain critical hub regions, resulting in a higher level of amyloid burden than their topological neighbors. However, current approaches for hub identification only focus on the topological structure of brain networks without considering the spatial distribution pattern of neuropathological burden residing within networks. In this work, we proposed a novel method for identifying neuropathological hubs that integrates both the neuropathological and topological information of brain networks based on multimodal neuroimages, where the removal of hubs will result in a maximum decomposition in brain networks as well as a minimum variation in neuropathological burdens. Experimental results on real datasets demonstrated that regions identified as neuropathological hubs suffer a greater risk of neuropathological damage than those of conventional approaches, supporting the consensus distribution between hub nodes and neuropathological burdens.