All publications

Conference paper / 2026

Hemodynamic Neural Field: Continuous Blood Flow Estimation in Vessel Geometries via Global Shape Conditioning

Simon Perrin, Patryk Rygiel, Amin Ranem, Bram de Wilde, Sébastien Levilly, Jelmer M. Wolterink, Harold Mouchère

Off-Grid · MICCAI 2026 · Camera-ready

DeepSDF encodes vessel shape and local features that condition a neural field predicting three blood-flow velocity components.
Hemo-NF conditions a continuous velocity field on global vessel shape and local features learned by DeepSDF. Figure 1 · Perrin et al. (2026) · Perrin et al. · official OffGrid MICCAI 2026 manuscript

A continuous representation of flow

Hemo-NF separates vessel representation from flow prediction. A DeepSDF model encodes the geometry, and a second neural field predicts velocity at spatial coordinates. The study compares ways of conditioning the flow field on global and local shape information.

Experiments use synthetic bifurcating arteries and examine both geometry reconstruction and velocity estimation. The official MICCAI record provides the camera-ready paper; final DOI and page information have not yet been assigned there.

Cite this work

@inproceedings{hemodynamicneuralfield2026,
  title = {Hemodynamic Neural Field: Continuous Blood Flow Estimation in Vessel Geometries via Global Shape Conditioning},
  author = { Simon Perrin and Patryk Rygiel and Amin Ranem and Bram de Wilde and Sébastien Levilly and Jelmer M. Wolterink and Harold Mouchère },
  year = {2026},
  booktitle = {Off-Grid · MICCAI 2026},
  url = { https://papers.miccai.org/miccai-2026-sat/Off_Grid_026.html }
}