Fitting a Multiple Source Photon Model for Monte Carlo Treatment Plan Verification
A Ventura1*, J Shih1, J Svoboda2, (1) Kaiser Permanente, Santa Clara, CA, (2) Kaiser Permanente, Roseville, CASU-E-T-19 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall
Purpose: To assess and reduce the difficulty of fitting a multiple source photon model for monte carlo treatment plan verification.
Methods: The EGS4 user code MCSIM, from Fox Chase Cancer Center, was chosen for its support of a multiple source photon model, of which the point and secondary (extrafocal) photon sources were utilized. A described method of fitting the secondary source to in-air output factors was implemented. Additionally, a method to fit the point source to a single large field dose distribution was explored. The point source fitter utilizes a database of pre-simulated mono-energetic fanlines to build distributions from arbitrary spectra. Perturbations are made to fanline spectra to reduce the errors along them. In this study the energy spectrum for each fanline has been limited to the log-normal distribution, which reduces the number of parameters for each to two.
Results: It was found that one spectral parameter could be set to a constant for all fanlines and the other restricted to linearity with respect to off-axis position. The model matched the outputs and distributions in non-superficial areas to within 2% for 6MV and 15MV Varian iX field sizes between 4 and 40 cm. Various types of treatment plans were then successfully verified, including 3D, VMAT, IMRT, and an iPlan Monte Carlo stereotactic lung to within 3% (tumor dose).
Conclusion: With such tools it is practical for a non-research physicist to fit a two source photon model for the purpose of monte carlo treatment plan verification. The only commissioning data needed are in-air output factors, a single large field dose distribution, and the usual machine parameters provided by LINAC vendors for clinical second check programs. Even when only photons are simulated and spectra are greatly simplified it is possible to achieve acceptable results for non-superficial tumors. Furthermore, this is achieved without proprietary machine specifications.