Premium
SU‐E‐T‐470: Monte Carlo Simulation of Fast Neutron Treatments for Head and Neck Cancers
Author(s) -
Moffitt G,
Stewart R,
Sandison G,
Emery R,
Jevremovic T
Publication year - 2013
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4814903
Subject(s) - monte carlo method , imaging phantom , physics , collimator , neutron , dosimetry , percentage depth dose curve , optics , nuclear medicine , photon , field size , computational physics , ionization chamber , nuclear physics , mathematics , medicine , ion , statistics , quantum mechanics , ionization
Purpose: To quantify the physical characteristics of a therapeutic fast neutron beam at the University of Washington (UW) using a Monte Carlo model validated against measured percentage depth‐dose (PDD) and lateral profiles. Methods: MCNPX modeling and simulation of the UW neutron therapy system's treatment head is divided into three stages. In Stage I, 50.5 MeV protons are directed at a Be target. Neutrons produced in the target pass through the primary collimator and flattening filter; photons and neutrons produced in Stage I are saved to a 100×100 cm 2 phase space file located just below the flattening filter. In Stage II, neutrons and photons are transported through the MLC and saved to a second phase space file. Automatic adjustment of the 40 individual MLC to create regular or irregular field shapes with or without a wedge is controlled by an in‐house developed python script. In Stage III, neutral particles are transported from the phase space file at the bottom of the MLC into a 40×40×40 cm 3 water phantom (150 cm source to surface distance). PDD curves and lateral dose profiles are computed from dose tallies in rectangular (0.1 cm 3 ) volumes. Results: PDD simulation for an open 10.4×10.3 cm 2 field agrees with measurement to within 4% up to 17 cm and within 8% up to 30 cm. A lateral dose profile at a depth of 1.7 cm matches well the overall shape of the profile. A representative simulation (10 11 source particles) requires about 150 hours of CPU time on a dual, 2.4 GHz 8‐core Intel Xeon server. Conclusion: The developed MCNPX model is capable of reproducing measured data with sufficient accuracy for clinical applications. Efforts to improve agreement in the penumbra of lateral profiles are underway as well as tests of the model for various sizes of open and wedged fields.