Source modeling for Monte Carlo dose calculation of CT examinations with a radiotherapy treatment planning system
Hans-Erik Källman1, Rickard Holmberg2, Jonas Andersson3
1Medical Radiation Sciences, Department of Immunology, Genetics and Pathology, Uppsala University, and Center for Clinical Research, County Dalarna, Bild och Funktionsmedicin, Falu lasarett, Falun SE-791 82, Sweden.
Purpose:
Radiation dose to patients undergoing examinations with Multislice Computed Tomography (MSCT) as well as Cone Beam Computed Tomography (CBCT) is a matter of concern. Risk management could benefit from efficient replace rational dose calculation tools. The paper aims to verify MSCT dose calculations using a Treatment Planning System (TPS) for radiotherapy and to evaluate four different variations of bow-tie filter characterizations for the beam model used in the dose calculations.
Methods:
A TPS (RayStation™, RaySearch Laboratories, Stockholm, Sweden) was configured to calculate dose from a MSCT (GE Healthcare, Wauwatosa, WI, USA). The x-ray beam was characterized in a stationary position the by measurements of the Half-Value Layer (HVL) in aluminum and kerma along the principal axes of the isocenter plane perpendicular to the beam. A Monte Carlo source model for the dose calculation was applied with four different variations on the beam-shaping bow-tie filter, taking into account the different degrees of HVL information but reconstructing the measured kerma distribution after the bow-tie filter by adjusting the photon sampling function. The resulting dose calculations were verified by comparison with measurements in solid water as well as in an anthropomorphic phantom.
Results:
The calculated depth dose in solid water as well as the relative dose profiles was in agreement with the corresponding measured values. Doses calculated in the anthropomorphic phantom in the range 26-55 mGy agreed with the corresponding thermo luminescence dosimeter (TLD) measurements. Deviations between measurements and calculations were of the order of the measurement uncertainties. There was no significant difference between the different variations on the bow-tie filter modeling.
Conclusions:
Under the assumption that the calculated kerma after the bow-tie filter replicates the measured kerma, the central specification of the HVL of the x-ray beam together with the kerma distribution can be used to characterize the beam. Thus, within the limits of the study, a flat bow-tie filter with an HVL specified by the vendor suffices to calculate the dose distribution. The TPS could be successfully configured to replicate the beam movement and intensity modulation of a spiral scan with dose modulation, on the basis of the specifications available in the metadata of the digital images and the log file of the CT.
