Abstract
BACKGROUND: Independent secondary dose calculation (ISDC) is becoming increasingly important for patient specific quality assurance. The most widely used analytical algorithms in ISDC are becoming challenged by Monte Carlo systems, which offer a potentially higher accuracy. PURPOSE: Quantify the benefit of Monte Carlo over analytical algorithms, and of customized beam models over generic beam models, in terms of clinically relevant parameters, action level, and workload. METHODS: A set of 100 patients across 20 case classes, all planned with Acuros XB (Siemens Healthineers) was analyzed with Mobius3D (M3D) (Siemens Healthineers) and SciMoCa (Radialogica LLC), both with custom beam models (SMCcbm) and generic beam models (SMCgen). Gamma pass rate (GPR) and mean target dose difference |ΔD| action levels were determined for various rates of QA failures. RESULTS: At a workload of < 10%, the action level for M3D was GPR (3%, 3 mm) < 90% and |ΔD| > 4.5%. For SMCgen, the action level was GPR (2%, 2 mm) < 95% and |ΔD| > 1.5%. For SMCcbm, it was GPR (2%, 1 mm) < 95% and |ΔD | > 1%. The combination of both criteria reduced the workload to < 5%. SMC failures could be traced back to differences in the patient density model of Acuros XB. Some M3D failures could be traced back to the handling of tissue heterogeneities. The different performance between SMCcbm and SMCgen was due to one (of three) generic beam models. CONCLUSION: Monte Carlo allows substantially stricter acceptance criteria and is sensitive enough to capture TPS commissioning errors. Generic beam models must be validated thoroughly before being put to use in ISDC.