Related Experiment Video
Updated: Aug 2, 2025
11:00
Experimental Investigation of Secondary Flow Structures Downstream of a Model Type IV Stent Failure in a 180° Curved Artery Test Section
Published on: July 19, 2016
11.7K
Machine learning and reduced order modelling for the simulation of braided stent deployment
Beatrice Bisighini1,2,3, Miquel Aguirre4,5,1, Marco Evangelos Biancolini3
1Mines Saint-Étienne, University Lyon, University Jean Monnet, INSERM, Saint-Étienne, France.
Frontiers in Physiology
|April 17, 2023
Summary
This study introduces a machine learning framework to predict the deployment of flow diverter stents for intracranial aneurysms. The model accurately forecasts successful stent placement, aiding minimally invasive aneurysm treatment.
Area of Science:
- Biomedical engineering
- Medical imaging
- Computational fluid dynamics
Background:
- Endoluminal reconstruction with flow diverters is a novel minimally invasive treatment for intracranial aneurysms.
- Predicting the deployed configuration of braided stents is challenging, and current imaging may be insufficient for treatment planning.
Purpose of the Study:
- To develop a fast and accurate machine learning framework to assist in planning and intervention for flow diverter stent deployment.
- To predict the success of stent deployment and approximate the final stent configuration.
Main Methods:
- A framework combining machine learning and reduced order modeling based on finite element simulations was developed.
- It includes a classification step for predicting deployment success and a regression step for approximating stent configuration using proper orthogonal decomposition and Gaussian process regression.
- The workflow was validated on an idealized intracranial artery model with a saccular aneurysm.
Main Results:
- Machine learning models achieved up to 95% accuracy in predicting deployment outcomes.
- The support vector machine model showed 93% accuracy and 97% specificity with a small dataset.
- Real-time predictions of stent configuration had an average error below the resolution of 3D rotational angiography (0.15 mm).
Conclusions:
- The developed framework enables rapid and accurate simulations of flow diverter stent deployment.
- This approach aids in treatment planning and interventional stages for intracranial aneurysms.
- The method retains mechanical realism and predictability of the deployed stent configuration.
