Development of An Accurate Monte Carlo Model of the Standard Clinac 6MV Beam
A Egan1,2*, W Laub2, (1) Oregon State University, Portland, OR, (2) Oregon Health & Science Univ, Portland, ORMO-D-108-5 Monday 2:00PM - 3:50PM Room: 108
Purpose: Monte Carlo (MC) models of radiotherapy beams can be used as a comparison benchmark for other dose calculation methods. In any such comparison process, the degree to which an identified deviation represents an actual error will depend upon the accuracy of the benchmark. As part of an ongoing dosimetric study, we have sought to model the standard Varian Clinac 6MV beam as accurately as possible with the BEAMnrc MC code.
Methods: In order to insure the reproducibility of our results, the accelerator head geometry was modeled exactly as specified by the manufacturer, with only the electron source parameters and jaw positions varied. Similarly, the widely available Eclipse 'Golden Beam' data was used for open field comparisons with a target matching criteria of 1% of local dose or 1 mm distance-to-agreement for all depths >= Dmax and field sizes ranging from 3x3 to 40x40 cm². Except for target bremsstrahlung settings, all other transport parameters were left at their default values. Simulations were run for varied monoenergetic electron energy [5.6:0.05:6.2]MeV, Gaussian intensity distribution FWHM [0.0:0.05:0.25]cm, and beam divergence [0.0:0.2:1.2]°.
Results: As has been previously established in the literature, matching was found to be strongly dependent on electron beam energy, intensity FWHM, and choice of bremsstrahlung settings. However, as matching approached the target criteria, results became increasingly sensitive to beam divergence as well. It was also found that when using the default bremsstrahlung settings, accuracy better than ~2% was unobtainable. Under the specified parameter resolution, matching of all points within 1.4%/1.0mm was achieved when using the NRC bremsstrahlung cross section, the higher termed Koch and Motz bremsstrahlung angular sampling, and electron beam parameters of 5.9MeV, 0.1cm, and 0.8°.
Conclusion: Under the listed constraints, significant improvements (e.g. matching ~ 1%/1mm) in Golden Beam modeling are not likely achievable.