Monte Carlo conversion for the Australian primary standard of absorbed dose to water in high energy photon beams

Radiotherapy treatment entails the delivery of large radiation doses to malignant tissues in the human body. These doses must be accurate in order to balance tumour control and damage to healthy tissues. The first step in accurate dosimetry is the calibration of radiation dosemeters by the national primary standards laboratory. Any uncertainties in this fundamental step will be passed on to every radiotherapy patient in Australia. Absorbed dose to water is the quantity used for the calibration of linear accelerator (linac) beams and many treatment planning systems. The work in this thesis is devoted to the establishment of the Australian primary standard of absorbed dose with clinically used high energy photon beams, and in particular to the Monte Carlo methods employed. The work described occurs in three stages: modelling of the accelerator head, modelling of the graphite calorimeter and water phantom in order to determine absorbed dose to water, and validation of the Australian primary standard of absorbed dose to water by comparison with international primary standards laboratories. The EGSnrc user codes BEAMnrc and DOSXYZnrc have been used for this work. The linac model is built using BEAMnrc component modules to match the components inside the real linac head. Validation of the linac model is performed by comparison of modelled PDDs and profiles with their measured counterparts. The ARPANSA measurement of absorbed dose to water is the basis for all absorbed dose calibrations performed in Australia. The determination of absorbed dose to water by ARPANSA begins with a measurement of absorbed dose to graphite. A graphite calorimeter is used to measure the heating caused by irradiation in order to determine the absorbed dose to graphite. The measured dose to graphite is converted to absorbed dose to water by a factor evaluated by Monte Carlo calculations. The conversion factor is calculated as the ratio of two components: the modelled dose to water at the reference depth in the absence of an ionisation chamber and the modelled dose to the core (the sensitive element) of the calorimeter. The calorimeter is modelled to replicate the device used with all Mylar coatings and air and vacuum gaps included. The physical calorimeter geometry is confirmed by kilovoltage imaging and gap corrections are calculated and compared to similar calorimeters in the literature for added confidence in the calorimeter model. The final stage of method validation involves comparisons with measurements performed by other researchers. Primarily this is done by comparing the determination of absorbed dose to water with other primary standards laboratories. This thesis presents a direct comparison performed in the ARPANSA linac beams and two indirect comparisons with measurements by the other participants completed at their respective laboratories. In all cases the ARPANSA measurement was lower than comparison participant. The difference between the ARPANSA measurement and that of the other participant was 0.02 to 0.46% at 6 MV, 0.41 to 0.76% at 10 MV and 0.68 to 0.80% at 18 MV. All results for the 6 MV beam agreed within 1σ. At 10 MV one measurement agreed within 1σ. The remaining 10 MV comparisons and all comparisons at 18 MV differed by between 1σ and 2σ. In addition to the validation methods, a detailed assessment of the uncertainties in the Monte Carlo conversion factor and the resulting calibration of an ionisation chamber are presented. The uncertainty in the calibration coefficient of an ionisation chamber after interpolation to the clinical beam energy is between 0.6 and 0.7%. The resulting quantity of absorbed dose to water is used to determine the calibration factor, ND,w [D, w subscript], of an ionisation chamber. The ratio of calibration factors measured in a linac beam and in ⁶⁰Co is the measured energy correction factor, kQ [Q subscript], at the linac beam quality. In addition to comparisons of absorbed dose to water, the measured kQ [Q subscript] values for commonly used ionisation chambers have been compared to measured and modelled values of kQ [Q subscript] published elsewhere. An important consideration in changing from using the IAEA kQ [Q subscript] values published in the TRS-398 Code of Practice to directly measured kQ [Q subscript] values at megavoltage energies is the shift caused in chamber ND,w [D, w subscript] factors. This varies with chamber type and beam quality. In this thesis four chamber types were considered: the NE 2571 Farmer chamber, and the NE 2611A, PTW 30013 and IBA FC65-G Farmer-type chambers. At 6 MV the expected shift in ND,w [D, w subscript] ranges from -0.2% to -0.9% across the four chamber types. For the 10 MV beam quality the expected shift is -0.8% to -1.3% and at 18 MV -1.1% to -1.4% is expected. The reason for these differences is twofold. The IAEA kQ [Q subscript] values are typically higher than measured kQ [Q subscript] values published by many authors. In addition to this, the ARPANSA measured kQ [Q subscript] values tend to be low compared to the average of many measured kQ [Q subscript] values. Regardless of the reasons, the shift has an impact on the beam calibration of clinical linacs and the implications of this effect are discussed.Thesis (M.Sc.(Med.Phy.)) -- University of Adelaide, School of Physical Sciences, 2015