Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Developing a contracting ventricle for patient-specific mitral valve simulations with fluid-structure interaction

Borén, Filip LU (2026) MVKM01 20261
Department of Energy Sciences
Abstract
Patient-specific fluid-structure interaction (FSI) simulations can improve the understanding of mitral valve dysfunction and support preoperative planning. However, existing contracting left
ventricular (LV) FSI models are often computationally demanding which limits their clinical applicability. This thesis addresses these limitations by developing a patient-specific contracting LV model within an existing left-heart FSI framework. In this original framework, referred to as the reference model, blood flow was driven through a fixed geometry using a mass flow
boundary condition applied at the ventricular apex. To introduce ventricular contraction into the framework, a regression-based morphing approach was implemented to enable... (More)
Patient-specific fluid-structure interaction (FSI) simulations can improve the understanding of mitral valve dysfunction and support preoperative planning. However, existing contracting left
ventricular (LV) FSI models are often computationally demanding which limits their clinical applicability. This thesis addresses these limitations by developing a patient-specific contracting LV model within an existing left-heart FSI framework. In this original framework, referred to as the reference model, blood flow was driven through a fixed geometry using a mass flow
boundary condition applied at the ventricular apex. To introduce ventricular contraction into the framework, a regression-based morphing approach was implemented to enable controlled, patient-specific LV deformation throughout the cardiac cycle. Furthermore, systolic pressure dynamics were improved by introducing an updated pressure outlet boundary condition. The implemented deformation model was evaluated against the reference model under the updated
boundary conditions using two patient-specific cases: MV6 and MV12.

The deformation model successfully reproduced overall LV volume dynamics, yielding end-systolic and end-diastolic volume deviations within 2% for MV6, and a mean systolic volume error of 5.5% for MV12 before numerical instabilities prematurely terminated the simulations. Under the updated boundary conditions, the reference model’s velocity accuracy degraded, showing errors of 25–30% against Echo Doppler data. In contrast, the deformation model demonstrated improved accuracy, bringing transmitral peak and average velocity errors below 15% for the MV6 case. Furthermore, allowing the fluid domain to develop within an enclosed, continuously deforming ventricle enabled the capture of cycle-dependent intraventricular flow structures that were impossible to reproduce with the original apex-driven approach. While the updated pressure outlet conditions eliminated artificial pressure-wave reflections, both models still generated unrealistically elevated systolic pressures. These findings demonstrate that a regression-based morphing approach is a computationally feasible alternative to complex FSI models, successfully capturing realistic intraventricular hemodynamics while preserving clinical applicability. However, future refinements to the outflow tract geometry and pressure boundaries remain necessary to further improve the model’s accuracy. (Less)
Popular Abstract
Patient-specific computer simulations could soon help doctors diagnose heart valve disease and test repair options before treatment begins. This thesis presents a new way of introducing the pumping function into these simulations to represent the beating heart.

Every year, millions of people worldwide suffer from valvular heart disease. While surgical repair can save lives, every patient’s heart is unique, making it highly challenging for doctors during preoperative planning to predict exactly how a specific valve repair option will perform. Advanced engineering computer simulations, known as Fluid-Structure Interaction (FSI), have the potential to transform this process. By coupling fluid dynamics with solid mechanics, these... (More)
Patient-specific computer simulations could soon help doctors diagnose heart valve disease and test repair options before treatment begins. This thesis presents a new way of introducing the pumping function into these simulations to represent the beating heart.

Every year, millions of people worldwide suffer from valvular heart disease. While surgical repair can save lives, every patient’s heart is unique, making it highly challenging for doctors during preoperative planning to predict exactly how a specific valve repair option will perform. Advanced engineering computer simulations, known as Fluid-Structure Interaction (FSI), have the potential to transform this process. By coupling fluid dynamics with solid mechanics, these simulations capture how flowing blood and deformable tissues constantly influence one another. These models can offer a deep, personalized understanding of a patient’s unique valve dysfunction. In the future, the ultimate goal is to allow clinicians to virtually evaluate different repair strategies during treatment planning to find the optimal solution for each individual.

A major hurdle for FSI models is their computational complexity. Previous models that included physical contraction of the heart were incredibly time-demanding, making them impractical for clinical use. To save time, an alternative approach is to keep the heart walls fixed and push blood through from the bottom of the ventricle. The problem? A stationary wall cannot replicate the complex, swirling fluid currents inside a beating heart, which may affect how the valve closes.

This thesis successfully bridges that gap. By implementing a "morphing" technique, a patient’s actual heart contractions—measured via standard, non-invasive clinical ultrasound—were mapped onto a simplified heart model. This allowed the virtual heart chamber to realistically contract and expand without requiring weeks of supercomputer calculations. Compared to the approach with stationary walls, introducing ventricular contractions improved the accuracy of the simulated blood flow velocity through the valve, while also successfully capturing the intraventricular swirling currents all within a clinically applicable time frame. This signifies an improvement to the model, and a further step towards clinical implementation of these simulations. (Less)
Please use this url to cite or link to this publication:
author
Borén, Filip LU
supervisor
organization
course
MVKM01 20261
year
type
H2 - Master's Degree (Two Years)
subject
report number
LUTMDN/TMHP-26/5674-SE
ISSN
0282-1990
language
English
id
9236176
date added to LUP
2026-06-16 11:52:06
date last changed
2026-06-16 11:52:06
@misc{9236176,
  abstract     = {{Patient-specific fluid-structure interaction (FSI) simulations can improve the understanding of mitral valve dysfunction and support preoperative planning. However, existing contracting left
ventricular (LV) FSI models are often computationally demanding which limits their clinical applicability. This thesis addresses these limitations by developing a patient-specific contracting LV model within an existing left-heart FSI framework. In this original framework, referred to as the reference model, blood flow was driven through a fixed geometry using a mass flow
boundary condition applied at the ventricular apex. To introduce ventricular contraction into the framework, a regression-based morphing approach was implemented to enable controlled, patient-specific LV deformation throughout the cardiac cycle. Furthermore, systolic pressure dynamics were improved by introducing an updated pressure outlet boundary condition. The implemented deformation model was evaluated against the reference model under the updated
boundary conditions using two patient-specific cases: MV6 and MV12.

The deformation model successfully reproduced overall LV volume dynamics, yielding end-systolic and end-diastolic volume deviations within 2% for MV6, and a mean systolic volume error of 5.5% for MV12 before numerical instabilities prematurely terminated the simulations. Under the updated boundary conditions, the reference model’s velocity accuracy degraded, showing errors of 25–30% against Echo Doppler data. In contrast, the deformation model demonstrated improved accuracy, bringing transmitral peak and average velocity errors below 15% for the MV6 case. Furthermore, allowing the fluid domain to develop within an enclosed, continuously deforming ventricle enabled the capture of cycle-dependent intraventricular flow structures that were impossible to reproduce with the original apex-driven approach. While the updated pressure outlet conditions eliminated artificial pressure-wave reflections, both models still generated unrealistically elevated systolic pressures. These findings demonstrate that a regression-based morphing approach is a computationally feasible alternative to complex FSI models, successfully capturing realistic intraventricular hemodynamics while preserving clinical applicability. However, future refinements to the outflow tract geometry and pressure boundaries remain necessary to further improve the model’s accuracy.}},
  author       = {{Borén, Filip}},
  issn         = {{0282-1990}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Developing a contracting ventricle for patient-specific mitral valve simulations with fluid-structure interaction}},
  year         = {{2026}},
}