Multiobjective robust optimization of coronary stents

• Published in: Materials & design Vol. 90; pp. 682 - 692
• Main Authors: Tammareddi, Sriram, Sun, Guangyong, Li, Qing
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 2016
• Subjects: Arteries; Coronary stents; Injuries; Mathematical analysis; Mathematical models; Multiobjective optimization; Nonlinear finite element analysis; Optimization; Robust design; Stents; Surgical implants; Surrogate modelling; Uncertainty; Coronary stents; Uncertainty; Nonlinear finite element analysis; Surrogate modelling; Multiobjective optimization; Robust design
• ISSN: 0264-1275; 1873-4197
• DOI: 10.1016/j.matdes.2015.10.153

Coronary stents are used extensively as a minimally invasive device for unblocking occluded coronary arteries stably. Restenosis is the re-occlusion of the artery post-stent implantation, which can be caused by an injury to the artery during stent deployment with neointimal growth on the metallic stent. The design of stent structure for minimizing the risk of arterial injury and maintaining stent stability involves several competing objectives. These objectives include reduction in dog-boning, foreshortening, elastic radial recoil, and stresses developed in the arterial wall during stent deployment. Another aspect is that the stent insertion and deployment process is subject to variabilities (uncertainties) such as slight movement of the stent on balloon catheter, and changes in stent material properties during manufacturing. Following the nonlinear finite element analyses of a parameterized stent model, in this study, the surrogate models are constructed to formulate the mathematical relationships between the stent geometrical parameters (control parameters) and biomechanical responses. With presence of uncertainties, the surrogate models include both mean and standard deviation components. To address the issue of stent design involving uncertainties, a multiobjective robust optimization is proposed here such that the effects of uncertainties on optimal objectives can be minimized. The Multiobjective Particle Swarm Optimization (MOPSO) algorithm is adopted to generate robust Pareto fronts for an optimal set of trade-offs between the objective functions while ensuring that the effects of the noise parameters are minimal. [Display omitted] •Parametrized coronary stent models with noise factors (design uncertainties) describing deployment behaviour was developed•Response surface method with Multiobjective Particle Swarm Optimization algorithm used to find optimal stent designs•Relative to the original stent design, thicker and wider stent struts and wider links give make stent design more robust