Scimoca™ Monte Carlo Benchmarking of the Cybercomm™ Reference Beam Data for Cyberknife® Commissioning
Abstract
Purpose
Reference beam data (RBD) is an efficient means to facilitate rapid and high quality linac and treatment planning system commissioning. We assessed the quality and reproducibility of the CyberComm™ RBD for CyberKnife® by creating a benchmark Monte Carlo (MC) beam model for the SciMoCa™ quality assurance (QA) product.
Methods
RBD was collected from 12 twinned CyberKnife linacs for all three collimator types (MLC, cone, variable cone), using the PTW MicroDiamond detector. The data was assessed and averaged with tolerances designed to meet clinical planning needs, guided by input from CyberKnife key opinion leaders. A MC virtual source model was created to match Depth-Dose-Curves (DDC) and output factors (OF) of the RBD (MLC). Beam models for the cone collimators were identical except for the collimator transport modules. Linac head and collimator geometries were derived from construction drawings. Agreement of MC simulations with RBD and customer measurements was assessed.
Results
The MC simulations reached an accuracy of a mean absolute deviation (MAD) of 0.23% for DDC of all MLC fields between 15.4 mm to 115.0 mm size, and a MAD of 0.17% for the OF, respectively. The average Gamma pass rate for dose deviation 0.5% / distance to agreement 0.5 mm for all cross-profiles was 99.6%. Agreement for the smallest fields < 10 mm was affected by small-field sensitivity of the MicroDiamond detector. The RBD-beam model matched the customer data well to within 20 keV deviation of the maximum energy. Small field OF (< 10 mm) were affected by variations of the focal spot radius in the order of 0.04 mm.
Conclusion
The CyberComm™ RBD shows very high internal consistency compared to a high-definition MC simulation. The RBD can be reproduced on customer linacs to an exacting standard, with minimal tweaking of the beam model only for small field OFs.