Automated VMAT Planning across TBI, CSI, and TMLI: Streamlining Workflow and Enhancing Dosimetric Quality
Abstract
Purpose
We developed a unified, fully automated, and open-source VMAT planning framework for three complex modalities: VMAT total body irradiation (VMAT-TBI), craniospinal irradiation (VMAT-CSI), and total marrow and lymphoid irradiation (VMAT-TMLI). The goal was to reduce planning time, improve consistency, and maintain or enhance plan quality. The tool is publicly shared via GitHub to support the global radiation oncology community.
Methods
An integrated scripting platform was built within the Varian Eclipse Scripting API. Plan quality was evaluated in 40 patients (10 VMAT-TBI, 20 VMAT-CSI, 10 VMAT-TMLI) by comparing automated plans with manually generated clinical plans using target and organ-at-risk DVH metrics and blinded physician review. Efficiency gains were estimated. Following validation, the tool was clinically adopted, and VMAT-TBI automation was incorporated into the multi-institutional COG ASCT2031 trial.
Results
For VMAT-TBI, auto-plans matched or surpassed manual plans in PTV coverage, global Dmax, and kidney sparing, with significantly reduced mean lung dose (5.4%±6.4%, p<0.05). Planning time decreased from 2–3 days to 3–5 hours, and 77% of plans were rated equivalent or superior. For VMAT-CSI, auto-plans reduced body V50% and mean doses to parotids, submandibular glands, and thyroid; 88.3% were rated equivalent or superior, with planning time reduced from ~5–6 hours to 1–2 hours. For VMAT-TMLI, auto-plans achieved equivalent target coverage and heterogeneity while significantly improving sparing of ovaries, kidneys, breast, larynx, thyroid, and oral cavity; reviewers preferred or rated auto-plans equivalent in 74% of cases.
Conclusion
A single automated scripting framework can effectively generate high-quality VMAT-TBI, VMAT-CSI, and VMAT-TMLI plans while substantially reducing planning time and improving consistency. The tool alleviates clinical workload and supports broader adoption of advanced radiotherapy. To date, it has been used clinically for 178 VMAT-TBI, 27 VMAT-CSI, and 1 VMAT-TMLI patients, with active open-source maintenance and dissemination.