Meeting Banner
Abstract #3665

Graph Diffusion

Ashish Raj1, Amy Kuceyeski, PhD1, Michael Weiner, MD2, Bruce Miller, MD3

1Department of Radiology, Weill Cornell Medical College, New York, NY, United States; 2Department of Radiology, University of California, San Francisco, San Francisco, CA, United States; 3Department of Neurology, University of California, San Francisco, San Francisco, CA, United States

Recent studies on prion-like proteopathy of dementias suggest transmission along fiber pathways of the brain network. In a recent paper we modeled the macroscopic consequences of network-centric propagation as a diffusion process on the structural (tractography-derived) brain network. The model accurately recapitulated known patterns of atrophy seen in several dementias. To our knowledge network diffusion constitutes the first fully quantitative, testable model of macroscopic transmission of degenerative processes in the brain. Its neurologic impact arises from its implication that diverse proteopathic etiologies could produce shared spatial patterns whose explanation requires neither selective vulnerability nor focal points of origin, nor differential stressor loads. Here we highlight this new advance and put it in the wider context of graph theoretic modeling of dementias. We believe that researchers involved in neuroimaging of neurodegenerative diseases can benefit from a deeper understanding of this exciting new technology.

Abstract PDF Presentation Video

Keywords

according accurately aging appears arises atrophic atrophy believe bench benefit best borne bottom brain clearly clinical close closely college common connectivity consequences constitutes content context correlation correspondence decomposition deeper degeneration dementia dementias depicted derived diagonal diamond diffusion diseased diseases disorders distinct diverse dots dynamic dynamics eigenvalue eigenvalues epilepsy etiologies evolves exciting explain explanation explore exponentiation fiber finally findings focal frost fully future give graph heat highlight homologies homologous hypothesis hypothesize implication implications implying individual investigations involved involving kind known labels leaving looking macroscopic makes match mathematical matrix measured medical merely might miller model modeled modeling models modes network networks neuron neuronal node outline overview paper papers particular patient patterns predict predicted prediction predictions predilection presentation prevalence process produce propagation provocative published quantitative radiology recent related requires selective simple solution spatial spatially spread strength structures studies subjects suggest summary support taken technology testable theoretic toxicity trans transmission understanding utility various vulnerability whose