CRCNS: TopoVess: A Topology-Infomred Vasculature Analysis Platform for Neuroscience

To advance our knowledge of brain functionality, it is essential to understand brain vasculature. Despite only
comprising 2% of the body weight, the brain consumes 20% of total body energy, which is supplied through
the vascular system. Local hemodynamics (the dynamics of blood flow) in the brain is tightly correlated with
the surrounding neuronal activity, continuously providing nutrients and bloodborne factors and eliminates
waste metabolites to regulate neuronal activity. The coupling between neuronal activity and blood flow,
termed neurovascular coupling, is linked to various physiological and pathological brain conditions. Chronic
neurovascular coupling dysfunction has been implicated as a key etiology underlying vascular diseases (e.g.,
stroke and vascular dementia), neurodegenerative diseases (e.g., Alzheimer's disease (AD) and Parkinson’s
Disease), sleep disorders and psychiatric disorders (e.g., autism). Many of these pathological conditions are
associated with learning deficits. With recent advance in imaging and tissue clearing techniques, we have
the opportunity to capture 3D vasculature of mouse brains in high quality. However, despite the high-
resolution images, the computational algorithm to analyze vasculature is far from satisfying. State-of-the-art
approaches are confronted by several challenges, such as limited generalizability across different data (e.g.,
different image modalities, different experimental settings, etc), prone to topological errors, lack of a powerful
statistical model.
To close the gap, we propose to establish a novel vasculature database and analytical platform for mouse
brain vasculature. Meanwhile, we will validate the proposed informatics tool with proof-of-concept studies of
neuroscientific questions and neurological disorders. A series of cutting-edge techniques are proposed in
this project. Developing topology-preserving vessel segmentation methods for a semi-automatic setting
(including semi-supervised segmentation, effective structural sampling for annotation with high uncertainty
and diversity) is highly innovative and critical to the study of vasculature. We also expect to develop novel
topological features, topological mappings and visual analytic tools for vasculature, based on advanced
mathematics of topology. Our proposed study will not only fill the knowledge gap by establishing a detailed
statistical mapping of vascular structures, but also provide a fundamental anatomical support for further
physiological and pathophysiological studies in understanding AD and stroke recovery/treatment.

Grant Number: 1R01NS143143-01
NIH Institute/Center: NIH
Principal Investigator: Chao Chen