Fast Monte Carlo simulation on a voxelized human phantom deformed to a patient.

ABSTRACT

PURPOSE: A method for performing fast simulations of absorbed dose using a patient's computerized tomography (CT) scan without explicitly relying on a calibration curve is presented. METHODS: The method is based on geometrical deformations performed on a standard voxelized human phantom. This involves spatially transforming the human phantom to align it with the patient CT image. Since the chemical composition and density of each voxel are given in the phantom data, a calibration curve is not used in the proposed method. For this study, the Monte Carlo (MC) code PENELOPE has been used as the simulation of reference. The results obtained with PENELOPE simulations are compared to those obtained with PENFAST and with the collapsed cone convolution algorithm implemented in a commercial treatment planning system. RESULTS: The comparisons of the absorbed doses calculated with the different algorithms on two patient CTs and the corresponding deformed phantoms show a maximum distance to agreement of 2 mm, and in general, the obtained absorbed dose distributions are compatible within the reached statistical uncertainty. The validity of the deformation method for a broad range of patients is shown using MC simulations in random density phantoms. A PENFAST simulation of a 6 MV photon beam impinging on a patient CT reaches 2% statistical uncertainty in the absorbed dose, in a 0.1 cm3 voxel along the central axis, in 10 min running on a single core of a 2.8 GHz CPU. CONCLUSIONS: The proposed method of the absorbed dose calculation in a deformed voxelized phantom allows for dosimetric studies in the geometry of a patient CT scan. This is due to the fact that the chemical composition and material density of the phantom are known. Furthermore, simulation using the phantom geometry can provide dosimetric information for each organ. The method can be used for quality assurance procedures. In relation to PENFAST, it is shown that a purely condensed-history algorithm (class I) can be used for absorbed dose estimation in patient CTs.