Sensitivity analysis of an asymmetric Monte Carlo beam model of a Siemens PRIMUS accelerator

Eric C Schreiber,Bruce A Faddegon,Daren L Sawkey

doi:10.1120/jacmp.v13i2.3402

Abstract

The assumption of cylindrical symmetry in radiotherapy accelerator models can pose a challenge for precise Monte Carlo modeling. This assumption makes it difficult to account for measured asymmetries in clinical dose distributions. We have performed a sensitivity study examining the effect of varying symmetric and asymmetric beam and geometric parameters of a Monte Carlo model for a Siemens PRIMUS accelerator. The accelerator and dose output were simulated using modified versions of BEAMnrc and DOSXYZnrc that allow lateral offsets of accelerator components and lateral and angular offsets for the incident electron beam. Dose distributions were studied for 40×40cm2 fields. The resulting dose distributions were analyzed for changes in flatness, symmetry, and off‐axis ratio (OAR). The electron beam parameters having the greatest effect on the resulting dose distributions were found to be electron energy and angle of incidence, as high as 5% for a 0.25° deflection. Electron spot size and lateral offset of the electron beam were found to have a smaller impact. Variations in photon target thickness were found to have a small effect. Small lateral offsets of the flattening filter caused significant variation to the OAR. In general, the greatest sensitivity to accelerator parameters could be observed for higher energies and off‐axis ratios closer to the central axis. Lateral and angular offsets of beam and accelerator components have strong effects on dose distributions, and should be included in any high‐accuracy beam model.PACS numbers: 87.55.K‐, 87.55.Gh

Highlights

33 Schreiber et al.: Sensitivity analysis of asymmetric beam model results to match experimental data should correspond to actual physical deviations from the initial model
Most radiotherapy computational models, including Monte Carlo, assume that non-beam– shaping accelerator components are cylindrically symmetric and aligned along the central axis. This assumption makes the geometry of the accelerator components easier to define, but does not always reflect the true configuration of a real accelerator.[2,3] The most obvious example of this is the small asymmetry in the in-plane direction attributed to a small angular deviation from the central axis of the electron beam as it emerges from the bending magnet
The accuracy of the accelerator simulation for a clinical accelerator with source and geometry asymmetry has been experimentally validated at 6 MV and 18 MV nominal X-ray beam energies with direct measurement.[3,4,8] The Monte Carlo code itself has been thoroughly benchmarked with measured data, and is among the most accurate codes for simulation of linac treatment heads used in radiotherapy (see, for example, Faddegon et al[9])

Summary

Introduction

33 Schreiber et al.: Sensitivity analysis of asymmetric beam model results to match experimental data should correspond to actual physical deviations from the initial model. In order to obtain the most accurate simulation results, the potential for asymmetric linac head components and off-axis electron source configurations should be included in the Monte Carlo commissioning process. Schreiber and Faddegon[2] performed a sensitivity study for large electron fields, including the effect of beam model asymmetries on the resulting dose distributions. The accuracy of the accelerator simulation for a clinical accelerator with source and geometry asymmetry has been experimentally validated at 6 MV and 18 MV nominal X-ray beam energies with direct measurement.[3,4,8] The Monte Carlo code itself has been thoroughly benchmarked with measured data, and is among the most accurate codes for simulation of linac treatment heads used in radiotherapy (see, for example, Faddegon et al[9]). The results of the sensitivity analysis, relating changes in source and geometry details to changes in dose distributions, are at least as accurate as the clinical beam comparison and experimental benchmarks

Materials and Methods

Findings

Conclusions