3042 - Automated Contouring, Planning, and Quality Assurance for VMAT Total Marrow and Lymphoid Irradiation with Initial Clinical Implementation
Presenter(s)
E. A. Simiele1, C. Hui2, I. O. Romero3, Z. Yang4, V. J. Heng5, L. Skinner6, L. Xing4, J. B. Ross7, R. T. Hoppe5, M. S. Binkley4, H. Elmariah6, S. M. Hiniker5, and N. Kovalchuk4; 1University of Alabama Birmingham School of Medicine, Birmingham, AL, 2Department of Radiation Oncology, University of California - Irvine, Orange, CA, 3Stanford University School of Medicine, Palo Alto, CA, 4Department of Radiation Oncology, Stanford University, Stanford, CA, 5Department of Radiation Oncology, Stanford University School of Medicine, Stanford, CA, 6Stanford University, Stanford, CA, 7Department of Radiation Oncology, Stanford Cancer Institute, Stanford, CA
Purpose/Objective(s): Total marrow and lymphoid irradiation (TMLI) enables target dose escalation with reduced exposure to surrounding organs at risk (OARs) compared with conventional approaches; however, its clinical implementation is complex and resource-intensive. To streamline and standardize delivery, we developed automated contouring, treatment planning, and physics plan-check workflows and made the scripts publicly available.
Materials/Methods: Fifty patients (age range 2–64 years) previously treated with volumetric modulated arc therapy total body irradiation (VMAT-TBI) were used to develop an auto-contouring model for relevant OAR segmentation. Auto-contours were evaluated using Dice Similarity Coefficient (DSC), 95% Hausdorff Distance (HD95), and qualitative physician review. An automated planning script was created using the Varian Eclipse™ API and tested in fifteen cases: five plans using 2 Gy TMLI, five using 12 Gy TMLI, and five using 20 Gy/12 Gy TMLI prescriptions (20 Gy delivered to bones, lymphoid, spleen and testes and 12 Gy to brain and liver). Dosimetric parameters, planning time, and blinded physician preference were used to compare automated and manual plans. Automated plan-preparation and physics plan-check workflows were implemented to improve efficiency, consistency, and safety, and the scripts were deployed clinically for treatment of the first patient receiving a 20 Gy/12 Gy TMLI prescription.
Results: TMLI target auto-contours achieved an average DSC of 0.89 ± 0.03, HD95 of 3.38 ± 1.46 mm, and reviewer ranking of 1.12 ± 0.06 (“acceptable-as-is”). Across all prescriptions (2 Gy, 12 Gy, and 20 Gy/12 Gy), automated VMAT-TMLI plans demonstrated comparable dosimetric quality to manual plans, with an average dose difference of -0.16% ± 7.00%. Five reviewers (four radiation oncologists and one medical physicist) selected 2 Gy and 12 Gy TMLI automated plans as equivalent or preferred in 74% of cases. Automated contouring and planning required 4-5 hours compared with an estimated 2–3 days for manual workflows. For 2 Gy and 12 Gy prescriptions, VMAT-TMLI achieved significantly greater OAR sparing compared with analogous VMAT-TBI plans, with an average dose reduction of -34.1% ± 9.4%. The oral cavity, lenses, eyes, and salivary glands showed the largest reductions (>50%, all p = 0.05). Clinical implementation demonstrated feasibility and safe delivery for the first patient treated with a 20 Gy/12 Gy TMLI regimen.
Conclusion: Automated contouring, planning, and plan-check workflows for VMAT-TMLI improved efficiency while maintaining clinically acceptable plan quality across multiple prescription levels. Successful clinical deployment supports broader adoption, and publicly available scripts may facilitate standardized multi-institutional implementation of TMLI and future cooperative group trials.