Monte Carlo verification of IMRT dose distributions from a commercial treatment planning optimization system

The purpose of this work was to use Monte Carlo simulations to verify the a ccuracy of the dose distributions from a commercial treatment planning opti mization system (Corvus, Nomos Corp., Sewickley, PA) for intensity-modulate d radiotherapy (IMRT). A Monte Carlo treatment planning system has been imp lemented clinically to improve and verify the accuracy of radiotherapy dose calculations. Further modifications to the system were made to compute the dose in a patient for multiple fixed-gantry IMRT fields. The dose distribu tions in the experimental phantoms and in the patients were calculated and used to verify the optimized treatment plans generated by the Corvus system . The Monte Carlo calculated IMRT dose distributions agreed with the measur ements to within 2% of the maximum dose for all the beam energies and field sizes for both the homogeneous and heterogeneous phantoms. The dose distri butions predicted by the Corvus system, which employs a finite-size pencil beam (FSPB) algorithm, agreed with the Monte Carlo simulations and measurem ents to within 4% in a cylindrical water phantom with various hypothetical target shapes. Discrepancies of more than 5% (relative to the prescribed ta rget dose) in the target region and over 20% in the critical structures wer e found in some IMRT patient calculations. The FSPB algorithm as implemente d in the Corvus system is adequate for homogeneous phantoms (such as prosta te) but may result in significant under- or over-estimation of the dose in some cases involving heterogeneities such as the air-tissue, lung-tissue an d tissue-bone interfaces.