Abstract
Hemodynamic biomarker estimation is crucial for the diagnosis and prognosis of cardiovascular diseases. Patient-specific computational fluid dynamics (CFD) simulations offer high-fidelity data but are computationally expensive and time-consuming, limiting their immediate implementation in resource-constrained clinical settings. To address this challenge, we propose Hemo-NF, a two-stage conditional neural field framework for continuous blood flow prediction in vessel geometries. Unlike state-of-the-art architectures whose complexity scales with explicit mesh size, Hemo-NF uses global shape conditioning to achieve constant processing complexity. It encodes vessel geometries using DeepSDF via an auto-decoding strategy and conditions a secondary neural field to predict three-dimensional blood flow velocity components. To bridge the gap between global shape representation and local geometric one, we propose to use locally extracted features from DeepSDF to condition the flow prediction network. Validated on a synthetic dataset of bifurcating arteries, Hemo-NF achieves competitive results compared to SEGNN when considering successful geometry encodings. We also provide a comprehensive evaluation of different conditioning mechanisms (concatenation vs. FiLM) and latent representations. This framework offers a promising approach to continuous, geometry-driven hemodynamic assessment.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/Off_Grid_026.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=vtETPTLLN1
BibTex
@InProceedings{PerSim_Hemodynamic_MICCAISAT2026,
author = { Perrin, Simon AND Rygiel, Patryk AND Ranem, Amin AND de Wilde, Bram AND Levilly, Sébastien AND Wolterink, Jelmer M. AND Mouchère, Harold},
title = { { Hemodynamic Neural Field: Continuous Blood Flow Estimation in Vessel Geometries via Global Shape Conditioning } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17277},
month = {pending},
page = {pending}
}