Computational Models of Fluid-Structure Interaction in Healthy and Diseased Aortic Valves
PRESENTING AUTHORS FULL NAME: Rajat Mittal
INSTITUTION: Johns Hopkins University
ADDITIONAL AUTHORS NAMES, AS TO BE PUBLISHED: Shantanu Bailoor and Jung Hee Seo
BACKGROUND/PURPOSE: Heart valve diseases (HVD) affect nearly 10 million people in the US, of which, aortic valve (AV) disorders are most prevalent. Often dysfunctional valves cannot be repaired and require prosthetic replacements, through surgical or transcatheter techniques. Heart valve prostheses have resulted in positive patient outcomes and mitigation of patient mortality, but are expensive and susceptible to several potentially fatal failure mechanisms. Owing to their high cost and the risk involved for patients, design optimization of these valves has received much attention. Challenges in assessing their in vivo performance have motivated in vitro and in silico models of transvalvular flow in the aorta. While recent computational studies have focused on developing detailed nonlinear finite-element AV representations, we demonstrate how simpler computational models enable numerical “experiments” that can provide insights into the hemodynamic performance of healthy and diseased AVs.
METHOD: We use canonical models representing the human ascending aorta and AV and an extensively validated fluid-structure interaction solver to simulate transvalvular flow. Valve leaflets are modeled using a lumped element model as thin membranes, whose motion is governed by a balance between hemodynamic pressure loading and leaflet elasticity.
RESULTS: We demonstrate how healthy tricuspid valves and pathological conditions such as aortic stenosis and bicuspid aortic valves can be reliably simulated using simple changes to model parameters. Employing physiological flow conditions, hemodynamic metrics like transvalvular gradient and aortic jet velocity are shown to be in good agreement with the degree of induced stenosis. Furthermore, we use the in-silico models to explore a novel modality for remote, wireless monitoring of transcatheter aortic valve performance. By measuring transvalvular pressure gradient at discrete locations along the aorta lumen and using signal processing techniques and machine-learning algorithms, we determine optimal sensor positioning for accurate retrospective and prospective prediction of leaflet status.
CONCLUSION: Our study demonstrates that simplified leaflet dynamics can accurately simulate AV dynamics while capturing essential flow characteristics of transvalvular hemodynamics. Variation in model parameters allow for the modeling of a variety of valve configurations and pathologies. Finally, discrete aortic pressure measurements can be exploited to accurately predict valve health.