SE(3) Symmetry Lets Graph Neural Networks Learn Arterial Velocity Estimation from Small Datasets

Julian Suk*, Christoph Brune, Jelmer M. Wolterink

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingChapterAcademicpeer-review

30 Downloads (Pure)

Abstract

Hemodynamic velocity fields in coronary arteries could be the basis of valuable biomarkers for diagnosis, prognosis and treatment planning in cardiovascular disease. Velocity fields are typically obtained from patient-specific 3D artery models via computational fluid dynamics (CFD). However, CFD simulation requires meticulous setup by experts and is time-intensive, which hinders large-scale acceptance in clinical practice. To address this, we propose graph neural networks (GNN) as an efficient black-box surrogate method to estimate 3D velocity fields mapped to the vertices of tetrahedral meshes of the artery lumen. We train these GNNs on synthetic artery models and CFD-based ground truth velocity fields. Once the GNN is trained, velocity estimates in a new and unseen artery can be obtained with 36-fold speed-up compared to CFD. We demonstrate how to construct an SE (3 ) -equivariant GNN that is independent of the spatial orientation of the input mesh and show how this reduces the necessary amount of training data compared to a baseline neural network.

Original languageEnglish
Title of host publicationFunctional Imaging and Modeling of the Heart
Subtitle of host publication12th International Conference, FIMH 2023, Lyon, France, June 19–22, 2023, Proceedings
EditorsOlivier Bernard, Patrick Clarysse, Nicolas Duchateau, Jacques Ohayon, Magalie Viallon
Pages445-454
Number of pages10
ISBN (Electronic)978-3-031-35302-4
DOIs
Publication statusPublished - 16 Jun 2023
Event12th International Conference on Functional Imaging and Modeling of the Heart, FIMH 2023 - Online, Lyon, France
Duration: 19 Jun 202322 Jun 2023
Conference number: 12
https://fimh2023.sciencesconf.org/

Conference

Conference12th International Conference on Functional Imaging and Modeling of the Heart, FIMH 2023
Abbreviated titleFIMH 2023
Country/TerritoryFrance
CityLyon
Period19/06/2322/06/23
Internet address

Keywords

  • 2023 OA procedure

Fingerprint

Dive into the research topics of 'SE(3) Symmetry Lets Graph Neural Networks Learn Arterial Velocity Estimation from Small Datasets'. Together they form a unique fingerprint.

Cite this