SU-E-T-557: Dosimetric Evaluation of Elekta-XiO Superposition Convolution Algorithm for Siemens MultiLeaf Collimators

Abstract

Purpose: Elekta-XiO treatment planning system(TPS) uses the same machine layout to simulate both two level jaw system Linacs like Siemens and three level jaw system like Varian Linacs. The purpose of this study is to perform a systematic investigation of the dose calculated from TPS using superposition convolution algorithm and compare that to the measurements for these two types of Linacs. Method and Materials: Siemens Primus and Artiste accelerators with multi-leaf collimators as lower jaws and Varian Trilogy accelerators with tertiary multileaf collimators were used in this study. A series of specially-designed T-shaped MLC fields with different field-aperture-opening-ratios(FAOR) and several clinically used 3DCRT MLC-blocked fields were calculated using the XiO superposition convolution and Clarkson methods. These plans were delivered on the respective Linacs and measurements were compared to the calculations. Several TMR based dose calculations were performed as independent checks on both superposition convolution and Clarkson methods. Monte Carlo simulations were performed to verify the dose calculation and to study the sources of scattering contribution. Results: A larger discrepancy was found between the dose calculated by XiO using superposition convolution algorithm and the dose measured for the Siemens Linacs than Varian Linacs. An average difference of 1.8% between Siemens and Varian Linacs was found for the situation of FAOR >0.33, possibly resulted from modeling the non-existant lower jaws for Siemens Linacs in the XiO convolution algorithm. TMR based dose calculations using a combined collimator scatter and phantom scatter factors with a XiO back-projected-blocked equivalent square(BES) can estimate the delivered dose accurately(0.12% of measured values) including extreme cases for Siemens(FAOR<0.33). Conclusions: A correction of magnitude of 1.8% should be applied to Linacs that use MLCs to replace a pair of jaws in the XiO superposition convolution algorithm when FAOR is >0.33. For FAOR<0.33, the algorithm cannot predict the delivered doses accurately.