Imaging Dose Optimisation For Advanced Radiotherapy

Abstract

Cone-beam computed tomography (CBCT) images are used in image-guided radiotherapy (IGRT) to track anatomical changes throughout treatment and to set up patients to ensure accurate delivery of therapeutic radiation at each treatment session. The radiation dose received from CBCT is frequently considered a barrier to its daily use in clinical practice. Particularly in children, there are concerns over any extra dose — even small amounts — due to their increased sensitivity to radiation and long life expectancy contributing to an increased risk of radiation-induced cancers. CBCT is sometimes substituted by 2D imaging in these cases, resulting in less accurate patient set-up and does not allow visualisation of full anatomy. Anatomy visualisation is crucial in proton beam therapy (PBT), with which children are more frequently being treated. Here, any change in anatomy that the treatment beams traverse can lead to large deviations in dose to the tumour and nearby healthy organs. Low-dose CBCT would potentially allow for more frequent imaging and hence more accurate treatments. Most studies investigating lower-dose CBCT must acquire extra images of patients or are based on unrealistic phantoms. This thesis aimed to develop methods to simulate and assess low-dose CBCT from patient images in order to make practical recommendations for its implementation. First, a methodology was developed to allow for the reconstruction of CBCT using extracted projection data from a commercial PBT treatment machine. The method corrected for physical effects on image quality such as variable exposure, filter and detector motion, and beam hardening to eliminate artefacts in the reconstructed CBCT images. This reconstruction process was then applied to simulate low-dose CBCT on the PBT system by adding noise to projection images. Set-up accuracy and observer perception were evaluated for the simulated images. These results showed that a reduction in dose to 10% of the standard level had little effect on CBCT-CT registration or observer-rated image quality, despite being close to the dose of 2D imaging. Finally, a method to simulate low-dose 4DCBCT was developed and used to assess tumour and respiratory motion at reduced doses. The study showed that, although consistency in automatic registration and motion measurement decreased slightly at lower doses, the accuracy was still clinically acceptable with a registration success rate of 99.9%. In conclusion, this thesis implemented, for the first time, a low-exposure simulation method for a commercial PBT system, and evaluated 4DCBCT with a similar method. The thesis showed that there is no need to compromise between imaging dose and accuracy, as CBCT at typical 2D imaging dose levels is as accurate as standard CBCT; similarly, 4DCBCT is accurate at dose levels of 3DCBCT.

Date of Award	1 Aug 2023
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Marianne Aznar (Supervisor), Simon Meara (Supervisor), Michael Taylor (Supervisor) & Marcel Van Herk (Supervisor)