Development of a Monte-Carlo based tool for dosimetry studies with spectral scanners

Abstract

This thesis reports on the implementation of methods to optimise the MARS scan settings and conduct a novel material-specific dosimetry simulation using the MARS small-bore scanner. This thesis was undertaken because no guideline exists on how to optimise MARS scan settings and conduct material-specific dose simulation. The MARS photon-counting spectral computed tomography (CT) scanner is at the stage of conducting the first MARS human clinical trials. Therefore, the need to perform the clinical trials with optimal MARS scan settings are crucial. Also, the MARS scanner provides material-specific information without knowing their radiation dose deposition. Hence, it is important to characterise their dose depositions and reduces the patient's exposure while maintaining the image quality.

This study aims to use the As Low As Reasonably Achievable (ALARA) principle of radiation protection to determine the absorbed dose for spectral imaging of small animals that have sufficient image quality and material differentiation to meet clinical needs. Custom-built Perspex phantoms were used to measure signal-to-noise ratio and spatial resolution, and to measure radiation dose using thermoluminescent dosimeters. A multicontrast calibration phantom was used to assess material identification. Small animal imaging and dosimetry were then performed to demonstrate the study aim. The results suggest that the energy resolving capability of photon-counting CT maintains diagnostically relevant image quality with high levels of material discrimination at reduced radiation dose.

The material-specific dosimetry methods were established through developing and implementing the Geant4 Application for Tomographic Emission (GATE) Monte Carlo (MC) simulation program based on the MARS small-bore scanner. The simulated MARS scanner was verified with physical measurements. Further, a method for preparing spectral attenuation CT data for three-dimensional MC dose simulation has been established. Lastly, a method for demonstrating the potential use of the spectral material image for MC dose simulation has also been developed. The methods employed in this thesis can also be applied to optimise MARS large-bore and body-parts scanners scan settings and build a strong basis for personalised dosimetry.