Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.695881
Title: Managing radiotherapy treatment trade-offs using multi-criteria optimisation and data envelopment analysis
Author: Lin, Kuan-Min
ISNI:       0000 0004 5991 4739
Awarding Body: Lancaster University
Current Institution: Lancaster University
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Techniques for managing trade-offs between tumour control and normal tissue sparing in radiotherapy treatment planning are reviewed and developed. Firstly, a quality control method based on data envelopment analysis is proposed. The method measures the improvement potential of a plan by comparing the plan to other reference plans. The method considers multiple criteria, including one that represents anatomical variations between patients. An application to prostate cases demonstrates the capability of the method in identifying plans with further improvement potential. A multi-criteria based planning technique that considers treatment delivery is then proposed. The method integrates column generation in the revised normal boundary intersection method, which projects a set of equidistant reference points onto the non-dominated set to form a representative set of non-dominated points. The delivery constraints are considered in the column generation process. Essentially, the method generates a set of deliverable plans featuring a range of treatment trade-offs. Demonstrated by a prostate case, the method generates near-optimal plans that can be delivered with dramatically lower total fluence than the optimal ones post-processed for treatment delivery constraints. Finally, a navigation method based on solving interactive multi-objective optimisation for a discrete set of plans is developed. The method sets the aspiration values for criteria as soft constraints, thus allowing the planner to freely express his/her preferences without causing infeasibility. Navigation is conducted on planner-defined clinical criteria, including the non-convex dose-volume criteria and treatment delivery time. Navigation steps on a prostate case are demonstrated with a prototype implementation. The prostate case shows that optimisation criteria may not correctly reflect plan quality and can mislead a planner to select a “sub-optimal” plan. Instead, using clinical criteria provides the most relevant measure of plan quality, hence allowing the planner to quickly identify the most preferable plan from a representative set.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.695881  DOI: Not available
Share: