New empirical formula for neutron dose level at the maze entrance of 15 MV medical accelerator facilities.

摘要

An easily applicable empirical formula was derived for use in the assessment of the photoneutron dose at the maze entrance of a 15 MV medical accelerator treatment room. The neutron dose equivalent rates around the Varian medical accelerator head calculated with the Monte Carlo code MCNPX were used as the source term in producing the base data. The dose equivalents were validated by measurements with bubble detectors. Irradiation geometry conditions expected to yield higher neutron dose rates in the maze were selected: a 20 x 20 cm2 irradiation field, gantry rotation plane parallel to the maze walls, and the photon beams directed to the opposite wall to the maze entrance. The neutron dose equivalents at the maze entrance were computed for 697 arbitrary single-bend maze configurations by extending the Monte Carlo calculations down to the maze entrance. Then, the empirical formula was derived by a multiple regression fit to the neutron dose equivalents at the maze entrance for all the different maze configurations. The goodness of the empirical formula was evaluated by applying it to seven operating medical accelerators of different makes. When the source terms were fixed, the neutron doses estimated from the authors' formula agreed better with the corresponding MCNPX simulations than the results of the Kersey method. In addition, compared with the Wu-McGinley formula, the authors' formula provided better estimates for the mazes with length longer than 8.5 m. There are, however, discrepancies between the measured dose rates and the estimated values from the authors' formula, particularly for the machines other than a Varian model. Further efforts are needed to characterize the neutron field at the maze entrance to reduce the discrepancies. Furthermore, neutron source terms for the machines other than a Varian model should be simulated or measured and incorporated into the formula for accurate extended application to a variety of models.