Placental blood-flow velocity quantification from diffusion MRI

Motivation: Estimating blood flow velocity in capillary beds lacking voxel-scale coherence is challenging with current imaging techniques. Altered blood flow in small-scale placental capillaries is a key factor affecting the health of pregnant women and their fetuses. Goal(s): Estimating fetoplacental blood velocity directly from diffusion MRI. Approach: We perform Monte Carlo dMRI simulations with perfusion in synthetic capillary systems, training a machine learning algorithm to recover parameters like flow velocity from generated signals. This trained algorithm then estimates these parameters from in-vivo diffusion MRI of human placentas. Results: Our approach estimates placental blood velocities and diffusivities, yielding values comparable with published data. Impact: Our approach estimates blood velocity from existing diffusion MRI data, reflecting placental blood flow conditions and potentially helping to diagnose diseases like pre-eclampsia. This approach could also be applied to organs where blood velocity is clinically significant.