Myocardial perfusion modelling: MR image processing and statistical inference

Yang, Yalei (2022) Myocardial perfusion modelling: MR image processing and statistical inference. PhD thesis, University of Glasgow.

Abstract

Coronary heart disease (CHD) is a major source of human mortality worldwide. Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI) is widely used as a non-invasive method to assess myocardial perfusion, which can be used to diagnose and detect myocardial ischaemia. The main aim of this thesis is to develop novel robust statistical classification models to detect the precise location of the ischaemia within the myocardium using DCE-MRI myocardial perfusion data. Firstly, the myocardial blood flow (MBF), a critical parameter adopted to quantify the degree of ischaemia within the myocardium, is estimated using semi-quantitative or quantitative methods. The MBF map is used as the input training data for the classification methods. Secondly, a Gaussian mixture model and its modifications, i.e. a mixture model incorporating spatial constraints, are applied to classify the myocardial tissues based on the MRI data. Markov random field priors are introduced to represent the spatial prior information. Thirdly, hierarchical Bayesian classification models are developed to combine the physiological model or the MRI meta data with spatial or spatio-temporal prior information to classify the myocardial tissues based on the MRI data. Furthermore, a polar projection method is developed to map the myocardium image to an annulus. The thesis concludes with an outlook on future work on how the methods developed in this PhD project can be extended to longitudinal analysis of myocardial perfusion DCE-MRI.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Colleges/Schools: College of Science and Engineering > School of Mathematics and Statistics
• Supervisor's Name: Husmeier, Prof. Dirk, Radjenovic, Dr. Aleksandra and Gao, Dr. Hao
• Date of Award: 2022
• Depositing User: Theses Team
• Unique ID: glathesis:2022-83057
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 02 Aug 2022 08:32
• Last Modified: 05 Jun 2023 08:23
• Thesis DOI: 10.5525/gla.thesis.83057
• URI: https://theses.gla.ac.uk/id/eprint/83057