Commissioning and validation of a Monte Carlo algorithm for spine stereotactic radiosurgery

Abstract

AbstractPurposeA 6FFF Monte Carlo (MC) dose calculation algorithm was commissioned for spine stereotactic radiosurgery (SRS). Model generation, validation, and ensuing model tuning are presented.MethodsThe model was generated using in‐air and in‐water commissioning measurements of field sizes between 10 and 400 mm2. Commissioning measurements were compared to simulated water tank MC calculations to validate output factors, percent depth doses (PDDs), profile sizes and penumbras. Previously treated Spine SRS patients were re‐optimized with the MC model to achieve clinically acceptable plans. Resulting plans were calculated on the StereoPHAN phantom and subsequently delivered to the microDiamond and SRSMapcheck to verify calculated dose accuracy. Model tuning was performed by adjusting the model's light field offset (LO) distance between physical and radiological positions of the MLCs, to improve field size and StereoPHAN calculation accuracy. Following tuning, plans were generated and delivered to an anthropomorphic 3D‐printed spine phantom featuring realistic bone anatomy, to validate heterogeneity corrections. Finally, plans were validated using polymer gel (VIPAR based formulation) measurements.ResultsCompared to open field measurements, MC calculated output factors and PDDs were within 2%, profile penumbra widths were within 1 mm, and field sizes were within 0.5 mm. Calculated point dose measurements in the StereoPHAN were within 0.26% ± 0.93% and −0.10% ± 1.37% for targets and spinal canals, respectively. Average SRSMapcheck per‐plan pass rates using a 2%/2 mm/10% threshold relative gamma analysis was 99.1% ± 0.89%. Adjusting LOs improved open field and patient‐specific dosimetric agreement. Anthropomorphic phantom measurements were within −1.29% ± 1.00% and 0.27% ± 1.36% of MC calculated for the vertebral body (target) and spinal canal, respectively. VIPAR gel measurements confirmed good dosimetric agreement near the target‐spine junction.ConclusionValidation of a MC algorithm for simple fields and complex SRS spine deliveries in homogeneous and heterogeneous phantoms has been performed. The MC algorithm has been released for clinical use.