Skin dosimetry in MRI-linacs using MOSFET measurements and Monte Carlo simulations

Integrating magnetic resonance imaging (MRI) with a linear accelerator has facilitated technological advancements in magnetic resonance-guided radiotherapy (MRgRT). Unlike computer tomography (CT)-based image-guided online adaptive radiotherapy, online MRgRT offers concurrent superior soft-tissue visualisation during treatment without exposing the patient to non-therapeutic imaging doses. In the presence of the magnetic field, Lorentz-force perturbation of charged particles is known to alter the dose distribution inside and outside the patient’s body. For an MRI-linac system where the magnetic field direction is parallel to the direction of the radiation beam, the fringe field of the MRI unit can focus electron contamination from the linac head, resulting in significant increases in skin dose at the patient/phantom surface. Therefore, it is crucial to thoroughly investigate the dose distribution at the surface of such a system and develop strategies to mitigate this issue. Similarly, for an MRI-linac system where the direction of the magnetic field and beam are perpendicular to one another, the behaviour of charged particles in the magnetic field is altered. Consequently, dose distributions within and outside the primary field, differ from those observed in conventional linacs. In this thesis, the skin dose of the Australian MRI-linac at a source-to-isocentre distance (SID) of 2.47 m has been studied both experimentally using a high-resolution metal-oxide-semiconductor field-effect transistor (MOSFET) detector, known as the MOSkin™, and with Geant4 Monte Carlo simulations that utilised high-resolution voxel scoring. The largest measured skin dose using the MOSkin™ detector was 337.5±6.5%, for a 16.7×16.9 cm2 field size and is expressed as a percentage of the dose deposited at dmax without the magnetic field. Geant4 Monte Carlo simulations were used to examine the efficacy of two techniques: (1) off-axis irradiation and (2) the use of an electron filter, to reduce the large and clinically unacceptable surface doses observed during irradiation with the Australian MRI-linac system. The simulation results demonstrate both techniques’ ability to reduce the surface dose. A substantial proportion of the electron contamination can be separated from the primary beam by shifting the field off-centre and away from the central magnetic axis of the MRI scanner. Electron contamination within the off-centre field leads to a nonstandard dose distribution at the surface. The non-uniform dose distribution at the surface is affected by the gap between the edge of the field and the central magnetic axis. For the Australian MRI-linac system, the range of electron contamination in water is approximately 2 cm. Consequently, the 2 cm depth dose distribution of an off-centre field simulated with and without the magnetic field are approximately equivalent. By introducing a 2 cm water-equivalent material placed at the treatment surface alongside the off-centre field, separated electron contamination can be attenuated and the tissue beneath sparred. Another approach that was studied to reduce the surface dose at the treatment site involved an electron filter. Positioning an electron filter above the treatment site serves to attenuate upstream-generated electron contamination. The effectiveness of a 2 cm thick water-equivalent electron filter has previously been studied using experimental measurements. Presently, beam data has been collected on the Australian MRI-linac, with the intention to use the block in situ for impending clinical trials. Expressed as a percentage of the dose deposited at dmax without the magnetic field, the 2 cm thick water-equivalent electron filter positioned 5 cm above the treatment surface produces a surface of approximately 111% at best. In contrast, a thin Sn electron filter placed at the same distance above the treatment surface is predicted to reduce the surface dose to approximately 89%.