Abstract

Mechanical heart valve replacement is the preferred alternative in younger patients with severe symptomatic aortic valve disease. However, a series of complications represented in thrombus and
pannus formation are associated with the bileaflet mechanical heart valves. This leads to risks of valve leaflet dysfunction (i.e., obstruction), which can be a life-threatening event. Therefore, there is an urge for a rapid and accurate diagnostic of the valve dysfunction. As a golden standard for valve function assessment, echocardiography may fail to assess the valve functionality. This leads to using another imaging modality such as computed tomography where the patient can be
exposed to X-ray radiations multiple times during the course of diagnosis and treatment. Thus, another radiation-free diagnostic tool is highly demanded in practice. This study explores a group of flow-based diagnostic parameters for five different scenarios of dysfunctional mechanical aortic valve that can be encountered clinically. Introducing such parameters comes in parallel with the recent access to obtaining cardiovascular flow fields clinically via techniques such as 4D-magnetic
resonance imaging or echocardiography- particle image velocimetry. Using an in vitro setup, physiological flow parameters have been obtained in a silicone model of the ascending aorta where
an actual bileaflet mechanical valve is placed. The instantaneous velocity field is obtained using particle image velocimetry while a blood analogue has been used as the working fluid. By analysing the normalized axial velocity profiles downstream the valve, a clear discrimination between all dysfunctional cases can be reached. Moreover, the effect of the induced flow disturbance by the
valve dysfunction is quantified on various critical parameters such as the viscous energy dissipation, the wall-shear stress and the accumulated viscous shear stresses on passively advected particles.
These parameters can provide a clear insight into the progression of valve obstruction, in other words, a slight valve obstruction may lead to a complete blockage. This thesis also explores the application of modal decomposition in order to deduce the flow coherent features associated with each dysfunctional scenario. The results show the ability of proper orthogonal decomposition derived metrics to differentiate between healthy and dysfunctional cases. Moreover, reduced-order
modeling has been thoroughly investigated not only for the velocity field but also for higher order flow characteristics. Through the reduced-order models, a significant reduction on the original size of the velocity fields can be reached. Such application can gain more interest in managing
the storage of such data. Another axis of the analysis focuses on the Lagrangian perspective of fluid transport which can be an essential tool to understand the effect of valve dysfunction on the blood transport inside the ascending aorta. Under this analysis, the particle residence time inside the ascending aorta is analyzed which can be used to identify the dysfunctional scenario being associated with higher blood stasis time. This analysis can suggest which scenario induces higher
risk complications such as clot formation. As the transport of blood is governed by the Lagrangian coherent structures (LCS), revealing them through a geometrical approach has, therefore, been reported here. Utilizing the extracted repelling and attracting LCS, a new technique for the evaluation of the highest accumulated shear stresses is applied along the Lagrangian trajectory of particles being released from these structures where the highest stretching occurs. Such novel
analysis targets specific locations in the flow that witness higher level of shear rate which can be generalized to evaluate the hemodynamic performance of newly developed heart assisting devices.
Furthermore, the induced non-laminar flow behavior by the valve dysfunction is analyzed using the time-frequency spectra of velocity signals at selected points in the ascending aorta. As the
revealed LCS via a geometrical approach identifies only the transport barriers in the flow (i.e., the barrier between flow regions with different dynamics), a more intuitive method to reveal the blood transport skeleton (i.e., the interior of the coherent structure) is required. Therefore, the final axis of this thesis explores the application of Lagrangian trajectory graphs which can be combined
with a spectral clustering algorithm in order to detect the
ow coherent regions. The identified coherent sets have a large inner mixing between its members, while mixing between different sets
is minimized. The identified coherent sets have different distributions, sizes and numbers for each dysfunctional scenario. Moreover, the identified coherent sets enclose representative coherent
structures in the flow such as the vortical structures. Furthermore, classical measures of the constructed graph such as the nodal degree and clustering coeffcient provide more insight into the mixing and agglomeration between particles which is critical to understand the blood transport and to assess the potential of thrombus formation.