Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Mesh-Informed Reduced Order Models for Aneurysm Rupture Risk Prediction

Giuseppe Alessio D'Inverno, Saeid Moradizadeh, Sajad Salavatidezfouli, Pasquale Claudio Africa, Gianluigi Rozza

TL;DR

The study tackles real-time rupture risk prediction for thoracic aortic aneurysms by marrying full-order CFD with a graph-based surrogate: a mesh-informed reduced-order model using Graph Neural Networks to predict $WSS$ and $OSI$ across growth stages. It uses a non-Newtonian Casson fluid in a time-dependent incompressible Navier–Stokes framework with transitional $k-\omega$ SST turbulence, solved on FV meshes, and trains GNNs on surface graphs derived from FV discretizations. Interpolation (25%/50%/100% vs 75% test) and extrapolation (25%/50%/75% to 100%) tasks show the Graph Transformer-based GNN accurately tracks peaks and general trends, with OSI errors up to $0.5$ in interpolation and as low as $0.08$ in extrapolation, and demonstrates clinically relevant agreement when applying a rupture-risk OSI threshold of $0.3$. This framework offers fast, mesh-agnostic predictions suitable for clinical decision support, while acknowledging limitations such as lack of SE3 equivariance and single-patient data, and proposing directions like Temporal GNNs and multi-patient validation for broader generalization.

Abstract

The complexity of the cardiovascular system needs to be accurately reproduced in order to promptly acknowledge health conditions; to this aim, advanced multifidelity and multiphysics numerical models are crucial. On one side, Full Order Models (FOMs) deliver accurate hemodynamic assessments, but their high computational demands hinder their real-time clinical application. In contrast, Reduced Order Models (ROMs) provide more efficient yet accurate solutions, essential for personalized healthcare and timely clinical decision-making. In this work, we explore the application of computational fluid dynamics (CFD) in cardiovascular medicine by integrating FOMs with ROMs for predicting the risk of aortic aneurysm growth and rupture. Wall Shear Stress (WSS) and the Oscillatory Shear Index (OSI), sampled at different growth stages of the thoracic aortic aneurysm, are predicted by means of Graph Neural Networks (GNNs). GNNs exploit the natural graph structure of the mesh obtained by the Finite Volume (FV) discretization, taking into account the spatial local information, regardless of the dimension of the input graph. Our experimental validation framework yields promising results, confirming our method as a valid alternative that overcomes the curse of dimensionality.

Mesh-Informed Reduced Order Models for Aneurysm Rupture Risk Prediction

TL;DR

The study tackles real-time rupture risk prediction for thoracic aortic aneurysms by marrying full-order CFD with a graph-based surrogate: a mesh-informed reduced-order model using Graph Neural Networks to predict and across growth stages. It uses a non-Newtonian Casson fluid in a time-dependent incompressible Navier–Stokes framework with transitional SST turbulence, solved on FV meshes, and trains GNNs on surface graphs derived from FV discretizations. Interpolation (25%/50%/100% vs 75% test) and extrapolation (25%/50%/75% to 100%) tasks show the Graph Transformer-based GNN accurately tracks peaks and general trends, with OSI errors up to in interpolation and as low as in extrapolation, and demonstrates clinically relevant agreement when applying a rupture-risk OSI threshold of . This framework offers fast, mesh-agnostic predictions suitable for clinical decision support, while acknowledging limitations such as lack of SE3 equivariance and single-patient data, and proposing directions like Temporal GNNs and multi-patient validation for broader generalization.

Abstract

The complexity of the cardiovascular system needs to be accurately reproduced in order to promptly acknowledge health conditions; to this aim, advanced multifidelity and multiphysics numerical models are crucial. On one side, Full Order Models (FOMs) deliver accurate hemodynamic assessments, but their high computational demands hinder their real-time clinical application. In contrast, Reduced Order Models (ROMs) provide more efficient yet accurate solutions, essential for personalized healthcare and timely clinical decision-making. In this work, we explore the application of computational fluid dynamics (CFD) in cardiovascular medicine by integrating FOMs with ROMs for predicting the risk of aortic aneurysm growth and rupture. Wall Shear Stress (WSS) and the Oscillatory Shear Index (OSI), sampled at different growth stages of the thoracic aortic aneurysm, are predicted by means of Graph Neural Networks (GNNs). GNNs exploit the natural graph structure of the mesh obtained by the Finite Volume (FV) discretization, taking into account the spatial local information, regardless of the dimension of the input graph. Our experimental validation framework yields promising results, confirming our method as a valid alternative that overcomes the curse of dimensionality.
Paper Structure (17 sections, 13 equations, 23 figures, 3 tables)

This paper contains 17 sections, 13 equations, 23 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Materials and methods
  3. FOM
  4. Mathematical Model
  5. Turbulence
  6. Numerical discretization
  7. 3D Model
  8. Generation of Scaled Aneurysm Models
  9. Snapshot collection
  10. Graph Neural Networks
  11. Results and discussion
  12. Experimental setup
  13. Ablation study on the number of layers
  14. Interpolation task: numerical results
  15. Extrapolation task: numerical results
  16. ...and 2 more sections

Figures (23)

  • Figure 1: Final geometry of TAA after clean-up.
  • Figure 2: Computational Mesh.
  • Figure 3: Temporal evolution of area-averaged WSS on the aneurysm wall for different meshes.
  • Figure 4: Temporal profiles for the inlet and outlet boundary conditions.
  • Figure 5: Scaled volumes.
  • ...and 18 more figures