An investigation into computational modelling strategies to predict the in vitro performance of aortic valve replacements
Date
2023-06-19Author
Whiting, Robert
Metadata
Show full item recordUsage
This item's downloads: 73 (view details)
Abstract
The development and design of effective aortic valve replacements (AVRs) presents significant
challenges given the structural and hemodynamic performance requirements that must be achieved.
Conventional design approaches require substantial amounts of experimental testing, involving
complex pulsatile flow rigs that mimic the conditions of the cardiac cycle while allowing for accurate
measurements of valve performance. While these experiments are extremely valuable, they are time consuming and require multiple iterations of physical prototypes during the testing phase.
Computational simulations could overcome these challenges, streamlining the design process and
optimising the structural and hydrodynamic performance of AVRs under development. The objective
of this thesis is to investigate the potential of computational modelling to predict hemodynamic and
structural performance of aortic valves through a combined experimental-computational approach. In
particular, the thesis investigates both finite element and fluid-structure interaction based approaches,
implemented through Abaqus commercial software, and assesses their potential to robustly predict both
the structural and hydrodynamic performance of AVRs.
In this thesis, a tri-leaflet polymeric AVR, which was developed at the University of Galway, formed
the basis for both experimental and computational work. All devices were manufactured in-house
through compression moulding and a series of experimental bench top studies were carried out using
an in vitro pulsatile flow rig to investigate the hydrodynamic performance of different valve designs
according to ISO 5840. In parallel, a finite element based computational framework was developed to
predict the systolic and diastolic configurations from in vitro testing and a range of surrogate parameters
were proposed to provide direct insight into the in vitro hydrodynamic performance. This approach was
then used to examine the effects of asymmetric and regional calcification patterns on the stenosed
hydrodynamics of the aortic valve. Finally, a fluid-structure interaction model was developed using the
Abaqus Coupled Eulerian Lagrangian (CEL) approach to predict the structural and hydrodynamic
performance of the AVRs and to investigate the potential to conduct in silico bench testing of valve
devices. To enable this, a bench top rig was designed and manufactured to enable detailed measurements
of leaflet deformation and serve as validation for the Fluid Structure Interaction (FSI) simulations
conducted.
Through in vitro testing and in silico modelling, it was found that three-dimensional finite element
modelling could be used as a predictor of the in vitro hydrodynamic performance of tri-leaflet aortic
valve implants. Specifically, several surrogate measures were identified through regression analysis,
whereby leaflet coaptation area, geometric orifice area and opening pressure were found to be suitable
indicators of experimental in vitro hydrodynamic parameters of regurgitant fraction, effective orifice
area and transvalvular pressure drop performance, respectively. This finite element framework was used
to show that asymmetric and non-uniform calcification of aortic valves had a distinct effect on the
predicted hydrodynamic performance, measured parameters and the indicated in vitro hydrodynamic
performance. In particular, it was found that asymmetric calcification coverage was highly detrimental
to the systolic Geometric Orifice Area (GOA), while symmetric calcification was actually more
detrimental to diastolic parameters of diastolic GOA and Leaflet Coaptation Area (LCA). Finally, it
was found that the Abaqus/CEL fluid-structure interaction approach could accurately predict
experimentally observed leaflet deformations under two-dimensional flow conditions. In predicting the
three-dimensional performance of a tri-leaflet valve, it was found that the computational model could
capture certain features of the experimental performance across both a bicuspid and tricuspid valve,
including peak systolic GOA, but failed to accurately capture bulk measures of performance that were
present over the loading cycle (e.g. Effective Orifice Area (EOA)). This thesis highlights the distinct
challenges in validating FSI-based models of structural and hydrodynamic performance of AVRs, while
provided much-needed experimental data to this community.