Dose verification of an IMRT treatment planning system with the BEAM EGS4-based Monte Carlo code.

Intensity modulated radiation therapy (IMRT) has been increasingly used in radiotherapy departments during the last several years. A major advantage of IMRT in comparison to traditional three-dimensional conformal radiotherapy is the higher capability in providing dose distributions that conform very tightly to the target even for very complex shapes such as, for instance, concave regions. This results in a significant sparing of adjacent normal tissues. Different types of algorithms are employed in the IMRT dose calculation, from the simple pencil beam method, such as the finite-size pencil beam algorithm, to the more sophisticated algorithms, such as the kernel-based convolution/superposition ones. With the latter ones, electronic disequilibrium and inhomogeneities are better dealt with in comparison to the correction-based models like pencil beam. Nevertheless, even these types of algorithms may have some approximations that can potentially affect the dose results, especially considering that in an IMRT plan small segments or beamlets may be present for which electronic disequilibrium and inhomogeneities effects are of paramount importance. The goal of this work was to determine the accuracy in monitor units (MU) and dose distribution calculation of the algorithm implemented in the commercial treatment planning system PINNACLE3 (P3), for two IMRT plans with 6 MV photon beams. This system is based on a convolution/superposition with the Collapsed Cone approximation algorithm. The "BEAM" Monte Carlo (MC) code was employed as a benchmark in comparing the MU calculation and the dose distribution of P3. The model used to calculate the MU, with the separation of collimator scatter from the phantom scatter, valid for broad beams, was verified for narrow and irregular segments. The attention was focused on the way P3 calculates output factors (OF). A difference of 8% compared to MC was found for a particularly narrow segment analyzed. A dependence of the results on field size was found. For the complete plan, the agreement of dose distribution and MU calculation with MC results (affected by a dose uncertainty less than 0.5%) is very good: the dose difference at isocenter is 2.1% (1 standard deviation) for a "Prostate" site and 2.9% (1 standard deviation) for the "Head and Neck" site.