Arterial dysfunction is a precursor to aging-related cognitive decline and dementia; a clinical biomarker of dysfunction is blood flow velocity. Specifically, the detection of slowed flow in the anterior choroidal arteries is valuable for the early treatment of hippocampal degeneration, the hallmark of Alzheimer’s disease. 3D time-of-flight (TOF) magnetic-resonance angiography data, which can be collected in as little as 4 minutes, is the most promising non-invasive way for arterial imaging. However, 3D TOF does not provide blood-flow velocity information and is thus of limited clinical value. The alternative method, 3D phase-contrast angiography, provides quantitative blood flow velocity but requires 20-30-minute scans and more complex computations, and is not feasible for routine use in patients. In this work, we propose to develop a deep-learning framework for generating quantitative arterial-velocity measurements from 4 minutes of conventional 3D TOF data. To that end, we will first measure blood-flow velocity in the anterior-choroidal arteries using phase-contrast MRA in healthy adults, which we will use to train a network along with 3D TOF data from the same individuals. This project could provide a proof-of-concept that breathes new value into existing 3D TOF data and facilitates the inclusion of TOF in more clinical studies of dementia. The SUDS Scholar will: Familiarize with existing deep-learning tool for quantitative blood-flow velocity measurement from functional MRI signals; Perform biophysical modeling to understand the relationship between TOF signal intensity and blood-flow velocity; and, Identify ways to adapt existing deep-learning tool for estimating blood-flow velocity from TOF signal.