UCSF

Alzheimer’s Disease (AD) is one of the most common forms of dementia, yet the exact mechanisms by which it operates and progresses through the brain are not completely known. To help identify and characterize the underlying pathways, or subnetworks, of AD, a network diffusion model is built based on the heat diffusion equation, grey matter volume atrophy measurements, and individual brain structural connectivity networks. Longitudinal neuroimaging data from individuals in various stages of the AD disease spectrum are used to seed the model as well as provide empirical end-atrophy measurements to assess overall model predictive value. The dominant brain networks facilitating AD progression in each patient are extracted using regression analysis and these principal networks are then clustered into distinct groups by applying dimensionality reduction and classification. These identified groups expose AD-specific subnetworks, each having a unique distribution of disease spread topology through various regions of the brain. Further anatomical evaluation of these subnetworks shows that the affected regions of interest coincide with and support long-standing results of previous research.