Medical College of Wisconsin Milwaukee, WI, United States
J. Garcia Alvarez, A. Tai, K. Kainz, L. Puckett, M. E. Shukla, F. Zhu, E. M. Gore, and E. S. Paulson; Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, WI
Purpose/Objective(s): The number of patients undergoing re-irradiation to the same or adjacent site has risen in recent years. Efficient workflows that account for variations in anatomy and fractionation schemes between treatment courses are crucial to produce accurate composite re-irradiation plans that maximize target doses while minimizing risk to organs at risk (OARs) in the high dose overlap region. This work represents our institution’s projected re-irradiation Special Physics Consult workflow, emphasizing Deformable Image Registration (DIR) based dose accumulation quality assurance (QA) procedures. Materials/
Methods: The workflow, implemented in commercially available software, is intended to perform DIR between the prior planning CT (ppCT) and re-irradiation planning CT (rpCT). Prior doses are mapped to the rpCT and converted to equivalent dose in 2 Gy fractions (EQD2) before accumulation. DIR accuracy in aligning the boundaries of relevant OARs is assessed visually by warping the contours from the rpCT to the ppCT and quantitatively by calculating Dice similarity coefficient (DSC) and mean distance to agreement (MDA). Jacobian Determinant (JD), Curl, and Inverse Consistency Error (ICE) maps are used to evaluate local volume changes, smoothness, and consistency of the deformation vector fields. QA metrics are compared to recommended ranges from the literature. A MIM extension was developed to estimate voxel-wise dose mapping uncertainties, providing upper and lower bounds for cumulative dose-volume parameters. A second MIM extension was developed to convert the accumulated EQD2 doses into a bias dose at the re-treatment fractionation scheme, which can then be imported into the treatment planning system to guide optimization of the re-irradiation plan. The workflow was retrospectively tested on ten thoracic cases. Results: OARs such as the esophagus, aorta, pulmonary artery, heart, and larynx showed clinically acceptable alignment with DSC and MDA (Average±StdDev), 0.83±0.09 and 2.02±0.67 mm, respectively. In contrast, structures like the brachial plexus exhibited less favorable DSC (0.37±0.07) and MDA (3.09±0.65 mm). JD, Curl, and ICE values among all retreatment-relevant OARs were 1.07±0.18, 0.24±0.10, and 1.43±0.80 mm, respectively, indicating physically plausible deformations. The estimated EQD2 uncertainty bounds ranged from 1% to 15% of the OAR accumulated metric, allowing the prediction of worst-case scenarios. Conclusion: The robustness analysis incorporated in this workflow contributes to ensuring patient safety when using DIR-based background doses to guide re-irradiation planning and highlights potential limitations in certain scenarios. This workflow standardizes the re-irradiation process and maximizes target doses while minimizing OAR composite dose in high-risk regions.
