Development of a Monte Carlo simulation method for the evaluation of dose distribution calculations of radiotherapy treatment planning systems

[摘要] In this study a method is described whereby the dose distributions calculated by anytreatment planning system (TPS) could be evaluated using dose distributions calculatedwith Monte Carlo simulations. The Monte Carlo dose simulations can be regarded as thegolden standard. The method developed in this study involved the Monte Carlosimulation of a Philips SL75/14 based generic accelerator with the BEAM code. This wasdone to obtain beam information stored in phase space files that were characteristic of thegeneric accelerator. This beam data were then used for the simulation of dosedistributions in a mathematical water phantom using the Monte Carlo code, DOSXYZ.The same beam data were used to generate the data base for the TPS that uses it for dosecalculations in CT based patient models. The BATHO and ETAR inhomogeneitycorrection algorithms implemented on a CADPLAN TPS were evaluated. The CT slices that make up the patient model, on the TPS, were transformed to materialdata. Each of these materials (57 in total) covered a discrete CT interval in a total CTnumber range of 3000 CT numbers. Each of the 57 materials was represented in thepreprocessor code (pEGS4) to allow dose simulations in realistic patient models with theDOSXYZ code. Dose distributions were calculated in a maxillary sinus (head), lung andprostate patient for photon beams with size 2x2, 5x5 and 10xlO cm2. These dosedistributions were calculated on the TPS using the BATHO and ET AR methods. TheDOSXYZ dose distributions were scaled to the TPS calculated dose distributions bynormalization to the dose in water at 2 cm depth on the beam central axis.Dose difference volume histograms, percentage depth dose curves and 2D dosedistributions were obtained to evaluate these dose distributions. The BATHO and ET ARmethods cannot model lateral and longitudinal electron transport through complex media.These effects were apparent in large inhomogeneities such as in the lung model where theMonte Carlo dose simulation gave a wider beam penumbra for the large field, and in thedeviation of the TPS dose distributions in these regions for the small field size. The method developed in this study could also be applied to any IPS that uses moresophisticated models. Manufacturers of IPS's in particular could use the methodsdescribed in this study to evaluate their dose calculation algorithms.Key words: Monte Carlo, CT based patient model, DOSXYZ, BEAM, Treatmentplanning, dose distributions, lung, maxillary sinus, water phantom, lateral electrontransport, TPS, inhomogeneity.

[发布机构] University of the Free State