ESTRO 2025 - Abstract Book

S3323

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

ESTRO 2025

Conclusion: Uncertainty estimation from a DL based segmentation model was seen to correlate with Dice values for segmentation of MR-guided radiotherapy prostate cancer images. This implies that uncertainty estimation could be used to label the quality of the segmentations in the online adaptive radiotherapy workflow. Keywords: uncertainty estimation, error detection References: [1] B. Segedin and P. Petric, “Uncertainties in target volume delineation in radiotherapy–are they relevant and what can we do about them?,” Radiology and oncology 50(3), 254 (2016). [doi:10.1515/raon-2016-0023].

3710

Digital Poster Quantifying intrafraction motion during cone-beam CT guided online adaptive radiation therapy in right sided whole breast irradiation Anna M. Dinkla 1,2 , Naomi E. Duits 3,1 , Angelique R.W. van Vlaenderen 1,2 , Koen J. Nelissen 1,2 , Karin N. Goudschaal 1,4 , Lars R.L. ter Beek 1 , Jessica H.I. van der Himst 1 , Jorrit Visser 1,2 , Judith G. Middelburg-van Rijn 1,2 , Desiree H.J.G. van den Bongard 1,2,5 1 Radiation Oncology, Amsterdam UMC location Vrije Universiteit Amsterdam, Amsterdam, Netherlands. 2 Cancer Treatment and Quality of Life, Cancer Center Amsterdam, Amsterdam, Netherlands. 3 Biomedical Technology & Physics, Vrije Universiteit Amsterdam, Amsterdam, Netherlands. 4 Radiation Oncology, Amsterdam UMC location University of Amsterdam, Amsterdam, Netherlands. 5 Cancer Biology and Immunology, Cancer Center Amsterdam, Amsterdam, Netherlands Purpose/Objective: We have implemented online adaptive radiotherapy (oART) for breast cancer. Since interfraction changes are eliminated in an oART workflow, current treatment margins might be reduced. However, due to the longer timeslots of oART workflow, intrafraction motion may be increased. This study aims to quantify intrafraction motion of the breast during oART. Material/Methods: The BREAST-ART trial (NL2021.0624) is a single-arm prospective study for breast cancer patients referred for postoperative RT 1 . The first 20 right-sided breast cancer patients treated on ETHOS (Varian, a Siemens Healthineers company, Palo Alto) were selected for this study between June 2022 and October 2023. The CTV included the whole breast, cropped 5 mm below the body contour. The prescribed dose was 5x5.2Gy to the CTV+5mm, delivered with 4-field IMRT. During the oART workflow, three CBCT scans were acquired; CBCT1 to check or adapt the propagated contours, CBCT2 to verify the target position pre-treatment and CBCT3 to check for intrafraction displacements (Figure 1). The time between CBCT1 and CBCT3 covers a time span of 15 minutes. A couch shift can be applied after CBCT2. CBCT1-CBCT2 and CBCT2-CBCT3 were rigidly registered using Elastix with a bounding box encompassing the CTV as region of interest. For the second registration, CBCT2 was corrected for the applied couch shift to measure patient motion only. Rotations were limited and centered around zero and therefore disregarded for this study. We recorded translations in 3 dimensions and tested for statistical significant differences of the translations (Wilcoxon Signed Rank test) between the first phase and the second phase of treatment.