Author: Alyssa Gadsby, William J. Godwin, Daniel G. McDonald, Jean L. Peng, Alek K. Rapchak, Sean A Roles, Austin M. Skinner, Stephanie Tan π¨βπ¬
Affiliation: Medical University of South Carolina π
Purpose: To characterize optimal optimization workflow and settings for a single large target utilizing the Leksell GammaPlan Lightning Optimizer.
Methods: An anonymized dataset containing a large cavity in the parietal lobe was used. 16 Gy, in one fraction, was prescribed to 95% of the PTV with a Ξ³-angle of 90Β° for each plan. Low-Dose and Beam-On Time (BOT) objectives were varied in increments of 0.1 (0 to 1) with the opposing objective set to 0 when not being examined. The process was repeated using a warm start method; an initial βbalancedβ plan, optimized with both objectives set to 0.5, was generated. Plans were analyzed for treatment time, gradient index (GI), and Paddick Conformity Index (PCI).
Results: BOT increased with the Low-Dose objective, with the largest increase occurring between Low-Dose settings of 0.5 and 0.7. BOT decreased with increasing BOT objective, substantially up to 0.7. PCI improved with increasing Low-Dose values up to 0.7 but plateaued thereafter. PCI worsened with increasing BOT objectives, particularly above 0.8. GI showed improvement with increasing Low-Dose values between 0 and 0.4 but did not improve at higher values. GI worsened with increasing BOT objectives up to 0.7 before stabilizing. Minimal differences between warm and cold start techniques were observed for BOT variations, though notable deviations in PCI occurred when Low-Dose penalties were adjusted. Cold start optimization slightly improved PCI (by 0.01), with no significant impact on other metrics.
Conclusion: Optimal Optimization settings should balance PCI, GI, and BOT. Low-Dose values above 0.6 did not significantly improve PCI or GI, while BOT continued to increase, suggesting an optimal value of 0.5β0.6. BOT values above 0.7 degraded PCI without improving BOT, indicating an optimal value of β€ 0.7. A lower BOT objective may improve GI. Cold start optimization improved PCI slightly but did not affect other metrics.