ESTRO 2025 - Abstract Book

S3296

Physics - Intra-fraction motion management and real-time adaptive radiotherapy

ESTRO 2025

(end-of-exhale; reference). In-house python scripts created deformed 4DCT phases for each 4DCT phase deformed to the 50% phase using DIR. To validate the deformation, the mean absolute error (MAE) was calculated, measuring average Hounsfield unit (HU) differences between each deformed 4DCT phase and the reference phase. MAE analysis focused on a region of interest (ROI), which included voxels covered by the treatment dose. This ROI was defined using a 20% prescription dose threshold and 2 cm expansion. DIR accuracy was further assessed using Dice Similarity Coefficient (DSC), Mean Distance to Agreement (MDA), and Hausdorff distance for bone structures (ribs, vertebrae). DSC measures contour overlap between segmented structures, MDA provides average positional differences between contours, and Hausdorff distance measures maximum contour deviations. Bone segmentation was independently performed on reference phase and two inhale phases (0% and 90%), which were then deformed to the reference using DIR (Figure1). Finally, 4D dose variations in organs at risk (OARs) were assessed by comparing the treatment plan on the reference phase to the same plan evaluated on the two inhale phases deformed to the reference using the respective DIR.

Results: The average-MAE within the ROI was 3 HU, showing minimal differences between deformed and reference phases. Bone contour accuracy for the two inhale phases deformed to the reference showed average DSC and MDA scores of 0.91 and 0.6 mm, meeting TG-132 tolerance levels (DSC>0.8;MDA<3mm). The average Hausdorff distance was 6 cm . Finally, dose evaluations showed minimal variations (mean=8cGy;SD=16cGy) in most OARs when comparing reference and inhale phases. One patient exhibited a mean dose increase exceeding 1 Gy to the heart and right atrium. Conclusion: DIR performance met TG-132 standards, achieving accurate contour deformation and minimal OAR dose variation in hemithorax. However, the dose difference to the heart was at maximum 103 cGy (mean=16cGy;SD=19cGy), which was likely due to DIR limitations in deforming high-motion areas like the lungs and diaphragm. References: Lena Nenoff et al. “Review and recommendations on deformable image registration uncertain- ties for radiotherapy applications”. In: Physics in Medicine Biology 68.24 (2023), 24TR01. doi: 10.1088/1361-6560/ad0d8a. url: https://dx.doi.org/10.1088/1361-6560/ad0d8a. Kristy K Brock et al. “Use of image registration and fusion algorithms and techniques in radiotherapy: Report of the AAPM Radiation Therapy Committee Task Group No. 132”. In: Medical physics 44.7 (2017), e43–e76. doi: 10.1002/mp.12256. Keywords: proton therapy, DIR, dose accumulation