Author: Chase Cochran, Shane McCarthy, Damodar Pokhrel, William St Clair 👨🔬
Affiliation: University of Kentucky, Department of Radiation Medicine, University of Kentucky, Radiation Medicine 🌍
Purpose: Manually generating intraocular stereotactic radiosurgery (SRS) plans involves significant challenges, including lengthy planning times and inter-planner variability. Knowledge-based SRS planning offers a novel implementation for intraocular malignancies, reducing planning times and ensuring consistent plan quality.
Methods: An in-house, clinically validated, HyperArc-based RapidPlan model for treatment of multi-lesion brain SRS was used to automate the planning of intraocular malignancies. Tumor delineation was done using high-resolution MRI with a 1-2 mm margin applied for PTV and prescribed a single-dose of 25 Gy. Treatment planning used HyperArc module on a TrueBeam LINAC (6MV-FFF) with Millennium 120MLCs. Dose calculation was performed using AcurosXB. Seventeen intraocular automated RapidPlan plans were compared against manually generated SRS plans for plan quality and deliverability. Data analysis was performed with p-value <0.05 indicating statistical significance.
Results: Compared to manual HyperArc plans, RapidPlan SRS demonstrated a similar dose-gradient but slightly lower plan conformality (p=0.001). It achieved a higher mean GTV dose (+1.3 Gy, p<0.001) and reduced maximum doses to the ipsilateral optic nerve (-2.1 Gy, p<0.001), lens (-4.8 Gy, p<0.001), and 50% of the eye (-2.7 Gy, p=0.035). Maximum brain dose increased (+1.8 Gy, p=0.011), with insignificant differences for brainstem (p=0.051) and cochlea (p=0.464). Treatment planning and beam-on times were reduced by ~165 minutes and 0.97 minutes (p=0.001), respectively. Patient-specific quality-assurance showed pass rates of 99.6% (manual plans) and 99.2% (RapidPlan) with 2%/2mm γ-criteria. Independent MC second checks agreed with AcurosXB algorithm within +/-5% for both plans.
Conclusion: HyperArc RapidPlan model used for intraocular SRS demonstrates significant advantages over manual SRS planning. RapidPlan plans showed a superior target dose; further sparing critical structures (optic nerve, lens, and ipsilateral eye) with slightly higher brain dose and reduced plan conformality. Further re-training this RapidPlan model by adding high-quality intraocular SRS plans could improve plan conformality and mitigate dose to normal brain.