Optimizing Sharpness: Impact of Control Ring Geometry on Dose Gradients In Varian Ethos 2.0 High-Fidelity Mode for Small Fields
Abstract
Purpose
Varian Ethos 2.0 recently introduced High-Fidelity (HF) Mode, featuring a 1.25 mm dose grid designed for small-target SRS/SBRT. While the Intelligent Optimization Engine (IOE) automates plan generation, the impact of user-defined optimization structures—specifically control rings—on maximizing dose fall-off remains unquantified. This study evaluates the relationship between control ring "gap" distance and the resulting dose gradient in HF mode.
Methods
Retrospective plans for targets ranging from 0.3 to 1.5 cm3 were optimized using Ethos 2.0 HF mode. Four strategies were compared: (1) Baseline IOE without rings, (2) HF-Mode with a 2 mm ring gap, (3) HF-Mode with a 1 mm gap (Aggressive), and (4) HF-Mode with a 5 mm gap (Offset). Plans were normalized to V_100%≈99% coverage. Evaluated metrics included the Paddick Gradient Index (GI), Conformity Index (CI), total Monitor Units (MU), and Modulation Complexity Score (MCS).
Results
Preliminary analysis of a pilot cohort indicates that optimization rings in HF-mode significantly improved both GI (15.9% mean reduction) and CI (40.7% mean improvement). For the 0.3 $cm^3$ target, CI improved by 50% (3.4 to 1.7), though GI remained high (>7.9), suggesting a physical penumbra floor for ultra-small volumes. Notably, while the 1 mm Aggressive strategy increased MU by 46.8% over baseline, the mean MCS improved from 0.747 to 0.872, suggesting HF-mode achieves superior gradients through more efficient, open-aperture sequencing rather than increased fragmentation.
Conclusion
Utilizing control ring structures in Ethos HF-mode optimizes dose fall-off and conformity without increasing plan complexity. The 1.25 mm grid allows for aggressive sculpting of the prescription isodose line, though physical beam characteristics limit gradient improvements for targets < 0.5 cm3. While these pilot results are promising, a larger patient cohort is required to establish statistical significance and further define the volume-dependent thresholds for gradient optimization.