Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.794185
Title: Automated MRI-based radiotherapy planning for brain tumours
Author: Johnstone, Emily Rose
Awarding Body: University of Leeds
Current Institution: University of Leeds
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Thesis embargoed until 01 Feb 2025
Access from Institution:
Abstract:
Aims: To investigate the clinical feasibility of performing magnetic resonance imaging (MRI)-only radiotherapy planning for patients undergoing treatment for brain tumours. Method: A systematic review was performed of methods which convert MRI scans into synthetic computed tomography (sCT) scans for the purposes of MRI-only radiotherapy planning. The feasibility of using two methods which had been developed at different centres for different anatomical sites, for the production of brain sCTs was assessed. The quality of automatically segmented brain structures generated using one of these techniques was determined. The temporal stability of geometric distortions of four MRI scanners over a year was investigated. A prototype of an MRI and CT compatible anthropomorphic head and neck phantom was developed for the purpose of quality assurance (QA). Results: The review demonstrated that atlas-based and voxel-based techniques were clinically useful methods for sCT generation, with machine learning techniques beginning to develop. An atlas-based technique (previously developed for prostates) and a neural network technique (not previously tested for brains) were found to produce sCTs of acceptable quality for the majority of metrics, after modifications to the clinical MRI scanning protocol were introduced. Dosimetric deviations meant that further optimisation would be required. Larger brain structures were generally well segmented automatically, whilst smaller structures were segmented erratically. The stability of MRI geometric distortions varied between scanners and sequences. A prototype anthropomorphic head and neck phantom was designed and constructed, with future modifications needed in certain areas. Conclusions: The feasibility of producing clinically acceptable brain sCTs along with automatically segmented structures, using models developed at different centres for different anatomical sites, has been shown. The variation in temporal stability of MRI geometric distortions means that careful consideration should be paid to the frequency of distortion QA. An MRI and CT compatible head and neck phantom prototype has been developed.
Supervisor: Henry, Ann ; Short, Susan ; Sebag-Montefiore, David Sponsor: Leeds Institute of Cancer and Pathology ; University of Leeds ; John Fisher Foundation
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.794185  DOI: Not available
Share: