A0822
Title: Topological network analysis of protein aggregates in Alzheimer's disease via bi-filtered persistent homology
Authors: Yuexuan Wu - University of South Carolina (United States) [presenting]
Kwun C G Chan - University of Washington (United States)
Abstract: Alzheimer's disease (AD) is characterized by the pathological accumulation of beta-amyloid and tau proteins in the brain. Understanding their spatial distribution and interplay is critical for uncovering disease mechanisms and progression. A novel topological framework is proposed for analyzing multimodal positron emission tomography (PET) imaging data to jointly capture beta-amyloid and tau deposition patterns and their interplay. Brain networks are constructed using partial correlations of standardized uptake value ratios across brain regions. A bi-filtered persistent homology approach is employed that enables threshold-free analysis of dual-channel network structures through topological invariants to capture multiscale features across modalities. For group comparisons, the networks are embedded into a hierarchical tree space via dendrogram construction and quantify group differences using geodesic distances in this non-Euclidean space. Applying the framework to two independent cohorts, region pairs consistently co-localized for beta-amyloid and tau are identified across datasets, and cross-cohort variability is revealed in mild cognitive impairment subgroups. The results also highlight tau's more complex aggregation behavior and its stronger association with clinical AD. These findings demonstrate the potential of topological network analysis and geometric representations to uncover biologically meaningful features from multimodal neuroimaging data.
