Stochastic Optimization and Uncertainty Quantification of Coronary Stent using Gaussian Process Model

Abstract

Coronary stents are small wire meshed tubes used to help restore blood flow in arteries blocked by a buildup of plague. Understanding the uncertainties in the manufacturing as well as operation of stents are vital in order to ensure performance of the device and more importantly the safety of the patient. By combining Finite Element Method (FEM) and Machine Learning (ML), a robust design to account for these uncertainties is made possible. In this study, the effects of geometry parameters along with operational parameters of the expansion of a Palmaz-Schatz stent model were studied. Simulations were conducted using an FEM model build with COMSOL Multiphysics^®, and the results were fed into ML software SmartUQ to build a surrogate model. Stochastic optimizations were performed accounting for uncertainties, and performance were compared to other literatures.