How robust are secondary dose algorithms for Patient-Specific quality assurance in modulated stereotactic Radiosurgery?

Abstract

Accurate dose verification is essential in stereotactic radiosurgery (SRS), where small fields and steep dose gradients may amplify uncertainties, especially with complex techniques like multitarget single-isocenter VMAT. This technical note evaluated the robustness of two secondary dose calculation algorithms for SRS patient-specific quality assurance (PSQA). RadCalc Collapsed Cone Convolution Superposition (RC-CCCS) and Monte Carlo (RC-MC) were commissioned on beam-matched VersaHD linacs. Percentage depth doses (PDDs), profiles, and output factors (OFs) of 6 MV FFF beams were compared with measurements to assess small-field modeling (1 × 1-10 × 10 cm2). Thirty VMAT treatments (57 brain metastases) were recalculated with RadCalc and benchmarked against Pinnacle3 TPS. Agreement was evaluated using local gamma analysis (2%/2 mm, 30% threshold), DVH metrics, and end-to-end tests. Both RadCalc algorithms showed strong agreement with measurements and TPS for PDDs, field sizes, and OFs. RC-CCCS enlarged the lateral penumbra, particularly in the in-plane direction (∼1.5 mm). In patient plans, significant differences were observed between algorithms (p < 0.001), with mean gamma passing rates of 96.8% for RC-MC and 94.1% for RC-CCCS. On average, RC-CCCS overestimated the central target dose by ∼ 5%. End-to-end measurements were highly consistent with both TPS and RC-MC (<1% mean deviation, p = 0.74), whereas RC-CCCS showed larger deviations (mean 5.7%,p < 0.001). RC-MC provides accurate and robust independent dose verification for intracranial SRS. RC-CCCS exhibited relevant inaccuracies in penumbra and 3D dose modeling, particularly in multi-lesion plans, highlighting the need to consider algorithm-specific limitations when implementing secondary dose calculations in PSQA workflows.