Retrospective dosimetric evaluation of the collapsed cone, AAA, and Acuros XB algorithms for lung cancer Halcyon VMAT plans

When RayStation is used for Halcyon treatment planning and the plan is transferred to the ARIA/Eclipse system for delivery verification, the dose must be recalculated using the Anisotropic Analytical Algorithm (AAA) or AcurosXB algorithm for compatibility. This study evaluated the dosimetric differences among the Collapsed Cone (CC), AAA, and AcurosXB algorithms for non-small cell lung cancer (NSCLC) volumetric modulated arc therapy (VMAT) plans on the Halcyon platform. Treatment plans for 60 lung cancer patients were initially generated using the CC algorithm in RayStation and then recalculated in Eclipse using AAA and AcurosXB without re-optimization or renormalization. Systematic variations were observed among the three algorithms. AcurosXB showed the largest reductions in target doses compared with CC (up to a 1.56% reduction in clinical target volume (CTV) D2%), while AAA demonstrated smaller differences. For planning target volume (PTV) metrics, both AAA and AcurosXB yielded lower doses than CC (AAA up to 2.16% in D95%; AcurosXB up to 1.58% in D2%). All variations in CTV and PTV metrics remained within approximately 1.7%. For organ-at-risk doses, AAA produced slightly lower values than CC, whereas AcurosXB yielded consistently lower doses across most parameters. Overall, this study shows that AAA and AcurosXB provide slightly lower dose estimates than CC for the same Halcyon plan, especially for PTV and organ-at-risk metrics. These results highlight the importance of consistent dose-calculation methodology in NSCLC radiotherapy, particularly in cross-platform workflows between RayStation and Eclipse.