Validation of Monte Carlo dose calculation algorithm for CyberKnife multileaf collimator.

PURPOSE To commission and evaluate the Monte Carlo (MC) dose calculation algorithm for the CyberKnife equipped with a multileaf collimator (MLC). METHODS We created a MC model for the MLC using an integrated module of the CyberKnife treatment planning software (TPS). Two parameters could be optimized: the maximum energy and the source full width at half-maximum (FWHM). The optimization was performed by minimizing the differences between the measured and the MC calculated tissue phantom ratios and profiles. MLC plans were calculated in the TPS with the MC algorithm and irradiated on different phantoms. The dose was measured using an A1SL ionization chamber and EBT3 Gafchromic films, and then compared to the TPS dose to obtain dose differences (ΔD). Finally, patient-specific quality assurances (QA) were performed with global gamma index criteria of 3%/1 mm. RESULTS The maximum energy and source FWHM showing the best agreement with measurements were 6.4 MeV and 1.8 mm. The output factors calculated with these parameters gave an agreement within ±1% with measurements. The ΔD showed that MC model systematically underestimated the dose with an average of -1.5% over all configurations tested. For depths deeper than 12 cm, the ΔD increased, up to -3.0% (maximum at 15.5 cm depth). CONCLUSIONS The MC model for MLC of CyberKnife is clinically acceptable but underestimates the delivered dose by an average of -1.5%. Therefore, we recommend using the MC algorithm with the MLC only in heterogeneous regions and for shallow-seated tumors.