Using Monte Carlo techniques to evaluate the dose distributions from a radiotherapy treatment planning system
English: In this study we used Monte Carlo techniques to simulate the SL25 linear accelerator treatment head using the BEAM Code. The main purpose of study was to evaluate the dose distributions obtained by the CADPLAN treatment planning system (TPS) for 8 MV photon beams of a SL25 linear accelerator in realistic patient models. Simulation of the treatment head involves modeling of the main components of the treatment head that have influence on the absorption and scattering of radiation. Simulation of the accelerator was done in two parts to minimize the simulation time. Analysis of the data generated by the BEAM code was carried out using BEAMDP, another subsidiary of the BEAM code. We calculated the beam characteristics which are difficult to measure experimentally, such as angular distributions, spectral distributions, planar fluence and planar energy fluence at a plane located just above the jaws of the treatment head. The phase space files at the isocenter were used as source input for DOSXYZ, a MC code to calculate 3D dose distributions in water or CT based phantoms. The DOSXYZ code was used to calculate depth dose and cross plane profiles in a water phantom. The data obtained with Monte Carlo methods were compared with that obtained by ionization chamber measurements. Depth dose and cross plane profiles obtained by Monte Carlo methods and ionization chamber measurements generally agreed within 2%. We created patient models from CT data of real patients using the CTCREATE option of the DOSXYZ program. Dose distributions for a number of field sizes and different anatomical sites were calculated with the DOSXYZ code and compared with corresponding dose distributions calculated by the TPS. The modified BATHO and ETAR inhomogeneity correction methods used in the TPS were evaluated. Results show that Monte Carlo methods can accurately reproduce ion chamber measurements in a water phantom. Monte Carlo techniques are very useful for evaluating the accuracy of dose distributions generated by treatment planning systems in patient based models where measurements are impossible. The BATHO and ETAR methods showed comparable results to the Monte Carlo results. This could be due to the inefficiency of the method (visualization of the dose distributions) that we used for the comparison of the results. A more quantitative method like the use of the dose difference volume histogram could give a more comprehensive evaluation.