Room: AAPM ePoster Library
Purpose: To compare an in-house GPU accelerated Monte Carlo dose simulation code ARCHER with TOPAS in the presence of magnetic fields in MRIgRT.
Methods: ARCHER is a next-generation MC code with computation platforms including CPUs, GPUs, and MICs. A magnetic field module was extended in the ARCHER. Dosimetric differences between ARCHER and TOPAS were calculated and compared in the presence of different magnetic field strength from zero up to 1.5 T orthogonally to the beam axis. Two irradiation situations were considered. First, 20-MeV electrons were injected perpendicularly to a water phantom. Then, 2-MeV photons were injected perpendicularly to a water phantom, a water-lung-water slab phantom, and a water-bone-water slab phantom, respectively. The phantoms surrounded by air. The dose distributions were scored using 3D grids comprising 0.2×0.2×0.2 cm3 voxels.
Results: For electrons in different magnetic fields, dose results show good agreements between TOPAS and ARCHER, with all of the 2%/2mm gamma pass rate being close to 100%. For photons, the maximum difference was found to be with the 1.5 T magnetic field. PDDs at the central axis and lateral dose profiles at different depths from two codes agreed for all points outside high gradient regions within 4.0% and 1.5%, 5% and 2%, 4.8%, and 1.7%, for water phantom, water-lung-water phantom, and water-bone-water phantom, respectively. The 2%/2mm gamma pass rate was greater than 98.9%, 98.3%, 98.5%, respectively. For the same model mentioned above, ARCHER can complete the dose calculations within 5 seconds running on the Nvidia GTX 1060 6G, while TOPAS would take about 100 minutes running on the Intel(R) Xeon(R) CPU E5-4607 using 48 threads.
Conclusion: ARCHER is an accurate and more efficient Monte Carlo method for X-ray dose calculation with magnetic fields compared with the reference TOPAS code.