Simulation of the TRIUMF Proton Therapy facility for applications to 3D printing in radiotherapy

Abstract

Proton therapy, a relatively young modality in radiation therapy, has proven useful in cases where a sharp dose gradient or low secondary irradiation is required. In Canada proton therapy it was performed at the TRIUMF Proton Therapy Facility in the treatment of large or difficultly positioned ocular melanomas. This rare primary malignant cancer of the eye has a poor prognosis if untreated. Patient vision sparing is critical for quality of life and is strongly affected by the accuracy of the chosen treatment. Reduction in irradiation of critical structures is a proven strength of proton therapy due to the high dose-gradient and finite range in tissue. But, with the advantage of steep dose gradients, comes the requirement of precision target positioning and planning. Monte Carlo particle transport software is a valuable tool for understanding treat- ment doses in cases where measurement is time consuming or difficult. Accurate simulation of primary proton dose to water aids in the evaluation of beam charac- teristics and allows for study into improving dose application for patient treatment. In this work, a full Monte Carlo model of the TRIUMF proton therapy facility was developed. Measurements were taken in water to validate simulated results within 2% over the treatment depth for a wide range of beam modulations. The second advantage of proton therapy lies in its reduced dose bath to healthy tissue. This is especially important in pediatric cases where extraneous dose comes with a high risk of secondary carcinogenesis. Whereas multi-angle photon treatments necessarily irradiate large volumes of healthy tissue to produce a flat target dose, proton treatments may irradiate a target with a single beam. With this advantage comes a trade-off - protons produce a large number of neutrons as they are prepared for patient treatment. These neutrons are the largest contributor to secondary dose in proton therapy and must be well modeled and shielded to ensure patient safety. The second part of this work involves the measurement of secondary neutron doses in the TRIUMF treatment room. Measurements were validated within 20% of simulated values with uncertainties dominated by calibration of the detector. Neutron doses to an anatomic human model showed that calibrated secondary doses were in line with similar treatment facilities reporting globally. Simulations indicated that the source of neutrons was primarily in the unshieldable region of the beamline opening. Thus the total treatment time was the determining factor in secondary dose to the patient. With primary proton dose well modeled, it became possible to study the pre- cision of treatment and possible avenues for improvement. The beam modulation wheels and optimization scheme was developed in the late 90‘s when computational and manufacturing technologies were less developed. Updated optimization methods indicated that moving to a smooth scheme of energy modulation, as opposed to a stepped modulation wheel, could improve distal dose sharpness. This was contrary to the long-held belief that there was an optimal number of steps for modulation. The third portion of this work explored the use of 3D printers to enable the fabri- cation of smoothly transitioning modulator wheels. Materials and printer methods were studied, indicating a strong candidate in the PolyJet TM method for beam mod- ulation. Both stepped and newly-optimized smooth modulator wheels were printed and validated. Total turnaround time for modulator production was under 24 hours - proving the feasibility of patient-specific beam modulation. The last portion of this work explored the use of positron emitting isotopes for dose validation. Protons traversing tissue or plastic generate β + emitting isotopes via nuclear interactions. The resulting back-to-back annihilation photons can be re- constructed into the isotope distribution produced by the beam. This can potentially provide information about beam position in the target and hence position of a phan- tom or patient. An anatomic 3D printed eye phantom was designed and irradiated to test the feasibility of this method. While a strong isotope signal was reconstructed, the test did not yield a viable technique due to the low resolution of the phantom scan. The phantom position was poorly reconstructed using the transmission scan. Despite this, it could be possible to improve this method by using other methods for phantom position registration.