ESTRO 2025 - Abstract Book

S2712

Physics - Dose calculation algorithms

ESTRO 2025

The table summarizes the ranges of differences between ion chamber and EPID measurements for fields from 3x3 to 30x30 cm 2 .

Conclusion: mGLAaS reading-to-dose conversion is accurate in both settings of specific or global configuration. A conversion field size specific shall be applied.

Keywords: EPID, reading-to-dose conversion

References: [1] Fogliata A, Liver JB, Reggiori G, Cozzi L. General aS1200 EPID reading-to-dose conversion. Radiother Oncol 2024; 194,Suppl1: S3267-S3269

4615

Digital Poster A fast Monte Carlo Algorithm optimized for dose monitoring in adaptive radiotherapy Bulent Aydogan 1 , Erik Pearson 1 , Carlos Bohorquez 2 , Christopher Stepaniak 1 , Jay Markham 2 , Melissa Cross 2 , Drew Bullock 2 1 Radiation and Cellular Oncology, University of Chicago, Chicago, USA. 2 Physics, LAP of America Laser Applications, L.L.C.,, Boynton Bay, USA Purpose/Objective: The next frontier in precision radiation therapy is online adaptive treatment. Monitoring and correcting for changes in the patient anatomy and/or motion occurring between simulation and beam-on for each fraction is one of the most significant remaining challenges in radiation therapy. One promising approach is surface guided radiation therapy (SGRT), i.e., multiple stereoscopic cameras for accurate real-time modeling of the patient surface, which can be used to model internal changes. A critical component for the success of SGRT is fast and precise dose calculation to provide the information needed for clinical decision support at the treatment console. This work evaluates accuracy and calculation time for a novel fast Monte Marlo (fMC) based calculation engine for an SGRT system in Luna 3D (LAP GmbH). (RaySearch Laboratories) with Monte Carlo. Lung plans were for 60Gy in 5 fractions according to NRG LU008 4 . Prostate plans were two-arc VMAT SBRT with ℞ of 36.25 Gy over 5 fractions, following SPARK trial guidelines 3 . Dose distributions were recomputed using Eclipse Acuros (Varian Medical Systems) and RadCalc’s (LAP GmbH) BEAMnrc implementation and the fMC algorithm. TheRadCalc BEAMnrc implementation is a CPU based statistical particle transport where the new fMC utilizes precomputed EGSnrc tracks and GPU acceleration. In RadCalc prostate plan computations were performed using the standard uncertainty level (~4.5%) and the low level (~2.5%) was used for the lung plans, defined at the 50% D max level. Material/Methods: From open access data sets, five non-small cell lung cancer 1 and five prostate 2 cases were replanned in RayStation

Resulting dose volumes were compared using gamma analysis criteria of 3% max dose and 2mm DTA, within external contour and ignoring values less than 10% of the ℞ dose, see Figure 1. Additionally, seven points are