Evaluation of the RadCalc collapsed cone dose calculation algorithm against measured data. | EN
Authors
Neil Richmond 1, Katherine Chester 2, Steven Manley 3
1Department of Radiotherapy Physics, Northern Centre for Cancer Care, Freeman Hospital, Newcastle upon Tyne, NE7 7DN, UK. Electronic address: neil.richmond@nhs.net.
2Department of Radiotherapy Physics, Northern Centre for Cancer Care, Freeman Hospital, Newcastle upon Tyne, NE7 7DN, UK.
3Department of Medical Physics, The James Cook University Hospital, Middlesbrough, TS4 3BW, UK.
Source
Medical Dosimetry: offical jounal of the American Association of Medical Dosimetrists, DOI: 10.1016/j.meddos.2023.04.004
Abstract
This work describes the experimental validation of the RadCalc (Lifeline software Inc, Tyler) collapsed cone dose calculation algorithm against measured data for a range of scenarios. 6 MV photon beam data were measured in a large water tank on a Varian TrueBeam linear accelerator. These were input into the RadCalc software, in conjunction with head geometry and output calibration information, then used to create a collapsed cone beam model. The model performance was assessed by comparison against measurement, using a selection of homogeneous and inhomogeneous geometries not incorporated into the original beam model. Dose calculations generated using the collapsed cone algorithm are generally in good agreement with measurement. However, the primary collimating of the linac is not accounted for in the RadCalc model and hence dose in the corners of large fields is significantly overestimated. Percentage depth doses were within 0.5% beyond a depth of 2 cm. The dose in the build-up region was underestimated by RadCalc Version 7.1.4.1, with (Distance To Agreement) discrepancies of up to 3 mm which were corrected in Version 7.2.2.0. Beam profiles for homogeneous phantom comparisons were within 2% in the central 80% of the field with out of field dose underestimated by no more than 3%. Dose comparisons in heterogeneous geometries were acceptable and generally within 3.5%. The largest observed differences were found at density interfaces and a result of the RadCalc dose resolution of 2 mm against 1 mm measured. Absolute dose comparisons demonstrated that RadCalc agreed with measurement to within 1.2% under homogeneous media irradiation geometries. For static beam IMRT deliveries agreement was within 2% or 2 mm of measured data, and for complex VMAT deliveries within 3% or 2 mm. The implementation of the (model-based) photon collapsed cone algorithm in RadCalc shows generally good agreement with measured data over a range of simple and complex scenarios considered.