Verification of a commercial treatment planning system based on Monte Carlo radiation dose calculations in intensity modulated radiation therapy

[摘要] English: Cancer treatment with external beam radiotherapy using the specialized technique of intensity modulation is a complex modality. The Treatment Planning System (TPS) is responsible for accurate calculation of dose to allow the radiotherapy team to make decisions on the patient treatment. The commercial TPS, XiO, utilizes a Multigrid Superposition algorithm as dose calculation engine, which is model based. Several approximations are inherent in this method. In-depth quality assurance (QA) of Intensity Modulated Radiation Therapy (IMRT) plans is necessary, and these tests are time-consuming and reduce the available clinical treatment time. Monte Carlo (MC) has been proven to be the most accurate method of radiation dose calculation. MC is a direct dose calculation method, and the EGSnrc codes are well suited for linear accelerator (linac) simulations.This study aims to be a first step towards full MC-based dose verification for IMRT dose distributions produced on XiO: developing the system and demonstrating the accuracy thereof.A generic virtual linac based on a typical Elekta linac was constructed using the EGSnrc MC software (BEAMnrc and DOSXYZnrc), for beam energies of 6 and 10 MV respectively. Simulations were either run on a watertank model or in air to produce beam data required for commissioning on XiO. Beam profiles, Percentage Depth Dose (PDD) curves and scatter factors for collimator and total scatter were extracted from the data. Software was developed to convert data to a format readable by the TPS. Modelling was done on XiO for all fields. A software graphical user interface (GUI) was developed to extract necessary information from dicom files required for MC calculations. This included CT data extracting and converting to EGSnrc format, reading all plan details, and creating scripts for automatic MC dose calculation execution. IMRT plans were created for 3 different treatment sites using the newly commissioned model on XiO. The modelling and simulation process was verified with MC dose calculations in scanned phantoms. After simulation, the IMRT plans were evaluated with isodose/profiles and 2D gamma analysis, as well as dose difference maps and Dose Volume Histogram (DVH) comparisons. The generic linac could successfully be created on BEAMnrc, and produced clinically acceptable beams. The data for commissioning was also generated successfully, and could be extracted and read into XiO after some de-noising filters were applied. Modelling on the TPS was done to an overall agreement level of 3%/3mm and 2%/2mm for small fields. Doses in the Prostate and Head-and-Neck IMRT plans compared well between XiO and MC for both energies. Gamma pass rates were above 90% for a criterion of 3%/2mm in a region of interest (ROI) covering the target and critical organs. Only slight overestimation of dose in bony regions was observed. The Esophagus IMRT plans however indicated some discrepancies in the dose calculation of XiO, especially in the low density regions, like lung. The 2D gamma pass rates were low, and DVH comparison indicated large overestimation of dose in the target volume, as well as in the Spine, as a direct consequence of errors in dose calculation of low density media.It is concluded that a dose verification system could successfully be developed for comparison of IMRT plans. Accurate modelling on the TPS was a vital step, and some possible issues were addressed. The system can be used routinely, and doses are calculated in a reasonable time with differences presented in a practical manner. The dose calculation of IMRT plans on XiO was compared to MC dose and found to be accurate for most treatment sites, independent of beam energy. However, caution is advised for cases where beams are directed through low density media, as clinically significant effects can possibly occur in patients.

[发布机构] University of the Free State