- April 26, 2019
- Category: Patient QA, Scientific Publications
Validation of EPID dose prediction and conversion models for flattening filter free beams
Achraf Ouakkad1, Maxime Goubert1, Laure Vieillevigne1, François Husson2, Laure Parent1
1 IUCT Oncopole, 1 avenue Irène Joliot-Curie,31059 TOULOUSE Cedex 9
2 Dosisoft, 45 Avenue Carnot, 94230 Cachan
Purpose: Electronic portal imaging devices (EPID) are interesting for pre-treatment quality assurance (QA) because of their high spatial resolution and ease of use. This study evaluated a new dosimetric portal method based on a superposition/convolution algorithm. It was tested for flattening filter free (FFF) photon beams.
Material and methods: Dosisoft EPIbeam software compares an image prediction generated from the DICOM RT plan and a portal image converted into a dose map at 5 cm depth in water using kernels to account for output factors, field penumbra and arm backscatter. Irradiations were performed with a Varian TrueBeam STx linear accelerator equipped with HD120 MLC and associated with aSi 1000 EPID. Dose prediction from RT plan and EPID image conversion models were assessed in 6 and 10 MV FFF beams by comparing the model to measurements. For output factor measurements, PTW 31010 0.125 cm3 ion chamber was used for output factors for 2×2 to 20×20 cm² field sizes at the isocentre. For clinical plans, prediction and conversion models were assessed with PTW 1000 SRS matrix (pixel resolution between 0.25 and 0.5 cm). Clinical plans were lung (6 MV FFF) and liver (10 MV FFF) stereotactic body radiotherapy plans using dynamic conformal arc technique.
Results: Predicted and converted output factors were within 2% of the measured values for field sizes between 2 and 20 cm². For clinical cases, comparison of dose prediction to matrix measurements gave an average gamma passing rate (2%-2.5 mm, global, 10% threshold) of (99.77±0.26)% and (99.98±0.04)% for 6 and 10 MV FFF beams respectively. Comparison of converted EPID image to matrix measurements gave an average gamma passing rate (2%-2.5 mm, global, 10% threshold) of (99.28±0.97)% and (99.98±0.04)% for 6 and 10 MV FFF beams respectively. Both prediction and EPID image conversion model are therefore validated for dynamic conformal arc technique. When the EPID image is used for pre-treatment QA, EPIbeam gave excellent gamma passing rates (2%-2mm, local, 10% threshold): for 6 MV FFF, the average pass rates were (98,79±0,61)% and for 10 MV FFF, the average pass rates were (98,55±0,47)%. Tolerance and action limits were calculated irrespective of the energy and were set to 96% and 87% respectively.
Conclusion: For field sizes between 2 and 20 cm², EPIbeam provided a good prediction of the dose in water at 5 cm depth and accurately converted the EPID image into a dose map in water. The software gave consistent results for the studied dynamic conformal arc clinical cases. This work should be extended to study more modulated beams, such as those used in volumetric modulated arctherapy and to study the sensitivity of the method to errors in delivery.