2741 - Development and Validation of an Automated Workflow for Re-Irradiation Constraint Generation In Radiation Therapy

Purpose/Objective(s): Re-irradiation is increasingly common due to improved cancer survival and expanded systemic therapy options. Published series report that 10–30% of patients receiving radiation therapy will undergo a second course. Re-irradiation planning is resource-intensive, requiring prior dose transfer, biologically adjusted calculations, and manual constraint derivation by physicians and physicists. Workflow variability increases inefficiency and risk of inconsistency. A safe, reproducible, and time-efficient method to generate re-irradiation constraints is needed to maintain treatment quality and patient safety. This study evaluates an automated MATLAB-based workflow compared with a standard manual clinical process.

Materials/Methods: Ten patients with abdominal malignancies treated with more than one course of radiation therapy were analyzed. The prior course CT was rigidly registered to the new planning CT, and the previous RT dose was transferred. The new CT, RT structures, and prior dose were exported for analysis. Two workflows were evaluated. Method 1 used a MATLAB script to import CT and RT structures and the prior dose to generate re-irradiation EQD2-adjusted physical dose constraints for each OAR (Stomach, Duodenum, Small and Large Bowel, and Spinal Cord). The script automatically created overlap structures between organs at risk (OARs) and a 20% isodose line from the prior prescription, providing a geographic constraint of critical regions requiring dose limitation. Method 2 followed a standard clinical workflow in which users manually extracted D0.1cc values for relevant OARs within the same 20% geographically constrained area, then entered them into a RayStation-based script to calculate adjusted dose constraints; overlap structures were manually generated. Processing time for both methods was calculated, and the resulting re-irradiation constraints for each OAR were compared. Paired t-test or Wilcoxon Signed-Rank testing was applied as appropriate.

Results: The automated workflow significantly reduced processing time (2 min [1–3] vs 16 min [12–20], p < 0.001). No statistically significant differences were observed between MATLAB and manual re-irradiation constraints. Mean absolute differences (Gy) were: small bowel 0.008 (0–0.86), large bowel 0.175 (0.01–1.22), stomach 0.535 (0–2.27), duodenum 0.02 (0–0.87), and spinal cord 0.027 (0.06–0.25). In one patient, an OAR was omitted during manual processing, highlighting potential user error.

Conclusion: An automated MATLAB-based re-irradiation workflow provides accurate and reproducible constraint generation with substantial time savings compared with manual methods. This approach maintains dosimetric equivalence while reducing workload and minimizing user-dependent variability, supporting safer and more efficient clinical implementation of re-irradiation planning.