Mai, Thanh Trung;
(2023)
Exploration of data-driven methods to reduce the impact of modelling choices on cardiovascular flow models.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
Peter_Thesis_Corrected.pdf - Accepted Version Download (53MB) | Preview |
Abstract
Computational modelling has shown to be a potential tool to revolutionise the prognosis and diagnosis of cardiovascular disease (CVD). However, the translation into a clinical setting has shown to be a significant challenge due to structural uncertainties which is the uncertainty stemming from modelling assumptions and choices in the model.\par In this thesis, the issue of structural uncertainty in the most modelling choices is identified, and reduced using a new modelling paradigm, data-driven modelling. Firstly, the most likely sources of uncertainty are identified through a sensitivity study of the most common assumptions employed in aortic flow simulations. Lumen area and wall boundary conditions are found to be the most likely sources of uncertainty. \par To address the uncertainties associated with boundary conditions, a deep learning method to model cardiovascular flow, the physics-informed neural network (PINN) is employed, which leverages partially known physics with partially known measurement to model the flow. PINN is found to be able to determine the velocity and pressure fields in aortic flows with good accuracy compared to CFD. Additionally, PINN learned the boundary conditions at the outlets and the wall from the training data, making this method agnostic to boundary conditions. The capabilities of PINN to model flows based on incomplete flow measurements can be potentially used to overcome issues with uncertainty in cardiovascular modelling. \par The capabilities of PINN was further explored through modelling several ill-posed problems and PIV measurements data. PINN has shown to have denoising properties and could infer other quantities from measured PIV data. This can further improve the quality of medical data. \par In summary, this thesis presents methodologies for identifying and overcoming uncertainties in cardiovascular modelling. As a result, these methods could be potentially used to model cardiovascular flow problems without introducing uncertainty into the models.\par
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Exploration of data-driven methods to reduce the impact of modelling choices on cardiovascular flow models |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2023. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Chemical Engineering |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10170566 |
Archive Staff Only
 |
View Item |
