Development and validation of an electron Monte Carlo model for the Elekta Infinity accelerator

Abstract

Purpose: The purpose of this work was to create a Monte Carlo BEAMnrc/EGSnrc model of the electron beam delivery system of an Elekta Infinity linear accelerator that could simulate electron percent depth doses and off-axis profiles to within a specified tolerance criteria, ±2% relative quantities (Dmax) or 1 mm distance to agreement (DTA), when compared to a subset of the measured commissioning data for the accelerator’s existing set of scattering foils and applicators. Methods: The model was constructed with the BEAMnrc user code from vendor provided schematics and correspondence with manufacturer engineers under a nondisclosure agreement. The spatial distribution of the initial electron source was derived by matching the large field (open-insert 25x25 cm2 applicator) cross beam profiles calculated by the model with the measured data. The measured data was re-sampled and symmetrized about the central axis (CAX) and the simulation data was smoothed before the two data sets were normalized to reference values along the CAX. The two data sets were compared to test to the hypothesis. Results: The model was able to meet the tolerance criteria (2% of Dmax or 1 mm DTA) for all large-field %DD curves. The model was unable to meet the criteria for any large field, shallow depth cross-beam profiles. The low energy (7-, 9-MeV) and high energy (16-, 20-MeV) models over-predict off-axis dose. The medium energy (10-, 11-, 13-MeV) beam model under-predict off-axis dose. These off-axis dose discrepancies were not pronounced at smaller field-sizes (6x6 and 14x14 cm2) for the subset of the clinical beam energies (7-, 13-, 20-MeV) that were compared. Conclusions: The Monte Carlo model of the Elekta Infinity electron beam delivery system could not simulate the considered subset of measured beam data to within the tolerance criteria of the study. The model could accurately reproduce the smaller field-sizes that were compared but could not reproduce the measured large field data. The large-field off-axis dose discrepancies observed with the medium beam energies (10-, 11-, 13-MeV) could be resolved by introducing a small initial angular spread (∼2°) to the incident electron source of the model.