SU‐E‐T‐680: Comparison between XVMC Monte Carlo and Eclipse AAA Dose Calculations for Rapidarc Plans

Purpose: To investigate the dose calculation accuracy of the Eclipse AAA dose calculation algorithm for Rapidarc treatment plans by means of recalculating Rapidarc plans with the fast Monte Carlo code XVMV.Methods: The fast Monte Carlo code XVMC was commissioned for a Trilogy linear accelerator (Varian, Palo Alto) with a 6x photon beam. Dose calculation tests with XVMC in solid water demonstrated the high accuracy of the beam model for different field sizes compared to measurements and calculations with the clinically commissioned AAA algorithm in Eclipse (Varian, Palo Alto). Dose calculations for conventional treatment plans in the absence of large areas of high or low density inhomogeneities for 3D conformal treatment plans, like 4‐field pelvis plans, also demonstrated a high level of agreement. Clinical Rapidarc treatment plans of prostate and H&N cases were recalculated with XVMC.Results: Comparison between AAA and XVMC dose calculations of Rapidarc plans show differences in the presence of inhomogeneities as they have already been discussed in a variety of publications comparing AAA or Collapsed Cone calculations with Monte Carlo results. In addition we found that the overall dose is overestimated in Eclipse by several Gray (see Fig. 2). A typical prostate or H&N plan will show a dose of 3–4 Gy less in the XVMC dose calculation compared to AAA calculations Conclusions: This work indicates that in addition to the well known limitations of AAA or Collapsed Cone algorithms with respect to how they handle and correct for lateral electron scatter and electron disequilibrium the accuracy of absolute dose calculations with AAA may also be limited in Rapidarc treatment planning. This can be explained with the extremely small islands of dose and the high percentage of area blocked by the MLCs as it can often be observed for control point of Rapidarc treatment plans. The author of this abstract holds a Varian research grant, is a guest speaker for Brainlab and has a part‐time commitment with Computerized Medical Systems.