Purpose: As prescribed by the IAEA-AAPM nonstandard beam dosimetry formalism, specialized radiotherapy treatment units which cannot achieve reference conditions established by conventional dosimetry protocols require detector quality correction factors. Monte Carlo is recognized as an essential method for the evaluation of these correction factors, but requires an accurate modeling and source parametrization of the radiotherapy device of interest. The purpose of this study is to propose a new maximum likelihood estimation method for the fitting of the source parameters of a CyberKnife M6 stereotactic radiotherapy unit.
Methods: Based on the approach of Francescon et al 2008, a formalism is developed to estimate the likelihood distributions of the energy and spot size of the electron beam incident on the target. A rigorous uncertainty budget, which includes detailed consideration of positioning uncertainties, is estimated. The EGSnrc code system is used to model: 1) the CyberKnife M6 unit and 2) four detectors recommended by the IAEA-AAPM protocol for relative dosimetry of small photon fields. Absorbed dose in the detectors’ sensitive volume are simulated for two setups, output factors and tissue-phantom ratios, using phase-space files generated with the CyberKnife model as the beam source. Corresponding measurements are performed on the CyberKnife M6 unit.
Results: The method is used to estimate the probability distribution functions of the electron energy and spot size. From these distributions, the parameters, and their respective uncertainties, are extracted using the solutions of maximum likelihood and the standard deviation of the obtained distributions. Results illustrate the agreement, using a coverage factor of 95%, between measured and simulated output factors and tissue-phantom ratios for most considered detectors and field sizes.
Conclusion: The proposed method allows an accurate beam parameterization of a CyberKnife M6 unit and a rigorous evaluation of uncertainties on the determined energy and spot size.