This is an archive of papers published by the staff and faculty of Fox Chase Cancer Center. For questions about content, please contact Talbot Research Library
Last updated on
Li JS , Shahine B , Fourkal E , Ma CM
A particle track-repeating algorithm for proton beam dose calculation
Physics in Medicine and Biology. 2005 Sp. Iss. SI MAR 7;50(5) :1001-1010
AbstractA particle track-repeating algorithm has been developed for proton beam dose calculation for radiotherapy. Monoenergetic protons with 250 MeV kinetic energy were simulated in an infinite water phantom using the GEANT3 Monte Carlo code. The changes in location, angle and energy for every transport step and the energy deposition along the track were recorded for the primary protons and all secondary particles. When calculating dose for a patient with a realistic proton beam, the pre-generated particle tracks were repeated in the patient geometry consisting of air, soft tissue and bone. The medium and density for each dose scoring voxel in the patient geometry were derived from patient CT data. The starting point, at which a proton track was repeated, was determined according to the incident proton energy. Thus, any protons with kinetic energy less than 250 MeV can be simulated. Based on the direction of the incident proton, the tracks were first rotated and for the subsequent steps, the scattering angles were simply repeated for air and soft tissue but adjusted properly based on the scattering power for bone. The particle step lengths were adjusted based on the density for air and soft tissue and also on the stopping powers for bone while keeping the energy deposition unchanged in each step. The difference in nuclear interactions and secondary particle generation between water and these materials was ignored. The algorithm has been validated by comparing the dose distributions in uniform water and layered heterogeneous phantoms with those calculated using the GEANT3 code for 120, 150, 180 and 250 MeV proton beams. The differences between them were within 2%. The new algorithm was about 13 times faster than the GEANT3 Monte Carlo code for a uniform phantom geometry and over 700 times faster for a heterogeneous phantom geometry.