- March 11, 2018
- Category: Patient QA, Scientific Publications
*Dosimetry improvement of an EPID-based pre-treatment verification software by the implementation of a dual photon beam model from linear accelerators
A. Rozesa, T. Billouxa, F. Hussonb, V. Dedieua
a Centre Jean Perrin, Clermont-Ferrand, France
b Medical Physics R&D, DOSIsoft, Cachan, France
Presented at SFPM 2018
Introduction: Point-kernel convolution-based dose calculation algorithms require an accurate modelling of the incident energy. The contribution of diffused photons by the primary collimator and the flattening filter (extra-focal source) is added to the primary source. Deviations up to 5% can be observed (Alaei, 2010) compared to the measurement for certain settings of collimation elements without an explicit modelling of the extra-focal source according to the position of the various collimation elements. EPID-based pre-treatment verification solutions generally uses fluence-based convolution models with the impulsion response of the detector. To be perfectly reliable, these methods must accurately predict the portal image to be compared with the acquired image. The aim of this work concerns the implementation and validation of an extra-focal source model for the computation of the predicted EPID image for standard (FF) and FFF (Flattening Filter Free) beams.
Methods: The linear accelerator used in this study is a TrueBeam STx® equipped with MLC HD120 and EPID as1200 with 6MV FF and FFF photons. Square fields with swapped openings of jaws and MLC leafs of 5 to 22 cm are designed with Eclipse® (Varian). Reference dose measurements in water are carried out on the central axis, with a microDiamond detector at SSD=95cm and 5cm deep. The predicted and acquired portal images are processed with the EPIbeam software (Dosisoft). The prediction model calculates the expected dose map at SSD=95cm and 5cm depth in water from the DICOM RT plan file, which is then compared to the acquired portal image converted to dose.
The accuracy of the TPS calculations, dose conversion and prediction models in EPIbeam, without and with implementation of an extra-focal source model including only the effects of the flattening filter, are confronted with the reference measures in water.
Results: The maximum discrepancy between the TPS calculations and the measurements is less than 1%. The dose values on the converted EPID image accurately match to microDiamond measurements (less than 1% of error). With the primary source model alone, the EPIbeam prediction model for FF photons shows dose differences up to 4% with the microDiamond detector with small MLC openings and retracted jaw positions. The observed deviations after the implementation of the extra-focal source model remain less than 1% for all collimations. For the FFF photons, deviations are in all cases less than 1%.
Conclusion: For FFF photons, the implementation of an extra-focal source model is optional since the contribution of the primary collimator appears negligible. However, this study confirms the importance of modelling the extra-focal source for FF photons to enhance the accuracy of the dosimetry calculations for pre-treatment checking.