Radiation source modeling for Monte Carlo based treatment planning systems.

摘要

In this study, we introduce a method to determine the energy spectrum delivered by a medical accelerator. The method relies on both Monte Carlo generated data and experimental measurements, but requires far fewer measurements than current attenuation-based methods, and much less information about the construction of the linear accelerator than full Monte Carlo based estimations, making it easy to perform in a clinical environment.; The basic model used in this work makes use of the quantum absorption efficiency concept, which gives the probability that a photon of energy $hn$ will deposit energy in a detector (film-screen detector in our case). Mathematically, our model is given by: M=Y0>T dYhn dhn Eavghne hndhn where M is the absorbed energy in the film-screen detector, $dYhndhn$ is the photon spectrum, $Eavghn$ is the average energy deposited per interacting photon, and $ehn$ is the quantum absorption efficiency, and $Y$ is the total photon fluence striking the detector. $ehn$ and $Eavghn$ were calculated by means of Monte Carlo simulation using the code MCNPX.; The method works as follows: first, the primary photon fluence exiting the target is calculated from first principles by dividing the target into thin slabs (50–100μm) and adding the bremsstrahlung contribution from each slab.; The electron fluence is calculated using the Phase Space Time Evolution Model, first proposed by Cordaro et al. and further refined by Huizenga et al. Ray tracing is used to attenuate the primary photon fluence as it passes through the flattening filter on its way to the detectors. Based on a detailed study of linear accelerator head scatter and of the known weaknesses of the Schiff cross-section we propose a multiplicative, energy-dependent empirical correction factor $fa,hn=expahn$ to take into account the head scatter energy fluence, where $a$ is a free parameter that is fixed by comparing the energy deposited in a screen-film detector irradiated by the spectrum in question to the theoretical prediction of the equation above. Since we do not know what the total fluence per monitor unit is striking the detector, we use two screen-film systems, with different quantum absorption efficiencies to determine the $a$ parameter. The detectors used are a 1mm copper plate attached to a Lanex Regular Gd₂O ₂S screen and a 1mm Aluminum plate attached to the same type of screen. These two detectors were characterized by means of Monte Carlo simulation.