Source model tuning for a 6 MV photon beam used in radiotherapy

The purpose of this work was to do a feasibility study on a procedure for tuning the mean energy T) and the spatial spread (sigma(R)) of the initial electron beam hitting the bremsstrahlung target as a part of a multicomponent source model, so that dose distributions predicted by Monte Carlo simulations (MC) will dosimetrically match measurements. The dose in a water phantom is considered to be a sum of doses produced by different components of a linear accelerator, namely target (TGT), flattening filter (1717), and primary collimator (PC). A histogram-based source model is used to model the phase space output of the 6 MV linear accelerator (VARIAN CL21EX). The phantom dose contribution from the target subsource has the greatest sensitivity to the initial beam parameters ((E) over bar sigma(R)), whereas the FF and PC subsource dose distributions axe only weakly dependent on them. In order to create a source model that could be tuned and used for a continuous range of (E) over bar and sigma(R), a method to interpolate across the histograms created with this data and to determine the sampling weights is developed, i.e. energy and radial-dependent interpolative polynomial fits to the histograms. The simulated dose distribution is then iteratively tuned to measured dose by changing the beam parameters, computing the corresponding dose distribution and compare simulation with measurements. This process is repeated until simulated dose matches measurements. The method ensures that the relative weights of radiation subsources are consistent with those that would be derived from full MC simulations of the treatment head. For comparison, the depth dose and the lateral dose profiles at various depths of the 10 x 10 cm(2) 20 x 20 cm(2), and 30 x 30 cm(2) field sizes are used. This study showed that a general source model tuning is feasible.