Physics-Informed Hemodynamic AI

Physics-informed learning is applied to anatomy-derived aortic hemodynamics and pressure-drop estimation in stenotic intracranial vessels.

Overview

Hemodynamic prediction is the central modeling task.

Flow-aware modeling bridges cardiovascular and neurovascular applications.

Imaging data and physics-informed computation are developed as complementary tools.

physics-informed deep learninghemodynamicspredictive modeling

Research overview

4D flow visualization representing AI-driven hemodynamic modeling.

Selected work

Anatomy-derived 3D Aortic Hemodynamics Using Fluid Physics-informed Deep Learning

Berhane H, Maroun A, Dushfunian D, Baraboo J, Martinez G, Jacobson T, Bagci U, Allen BD, Markl M. Radiology. 2025.

Non-invasive quantification of pressure drops in stenotic intracranial vessels: using deep learning-enhanced 4D flow MRI to characterize the regional haemodynamics of the pulsing brain

El Ahmar A, Schnell S, Ansari SA, Abdalla RN, Vali A, Aristova M, Markl M, Winter P, Marlevi D. Interface Focus. 2025.

Uncertainty-aware automated labeling of intracranial arteries using deep learning

Bisbal J, Winter P, Jofre S, Ponce A, Ansari SA, Abdalla R, Markl M, Welin Odeback O, Uribe S, Tejos C, Sotelo J, Schnell S, Marlevi D. BMC Med Imaging. 2026.