ESTRO 2025 - Abstract Book

S2065

Clinical - Urology

ESTRO 2025

automated segmentation approach was adopted for bladder delineation. Initially, a Cycle Generative Adversarial Network (CycleGAN) converted low-quality CBCT into high-quality pseudo-CT images while maintaining anatomical consistency. Then, a U-Net segmentation network delineated the bladder on the pseudo-CT images, followed by manual corrections by physicists. By rigidly registering the CBCT to the planning CT, bladder dose statistics were recalculated on dose files derived from the planning CT. Changes in bladder volume and mean dose for each CBCT fraction were quantified using Slicer software. A deep convolutional neural network was developed to explore whether dose variations could be predicted based on bladder volume changes. Results: Bladder DVH curves were obtained for each CBCT fraction by combining bladder contours and original treatment plans. Changes in bladder volume were categorized into five levels relative to the planning CT: <20%, 20 – 40%, 40 – 60%, 60 – 80%, and >80%, corresponding to 90 fractions (30%), 79 fractions (26%), 50 fractions (17%), 28 fractions (9%), and 53 fractions (18%), respectively. Mean dose variations followed similar grading: 203 fractions (67%), 55 fractions (18%), 11 fractions (4%), 11 fractions (4%), and 20 fractions (6%). Using the Mann-Whitney U test, no significant differences were observed in volume and dose variations between the two centers (p = 0.25 and p = 0.21, respectively). The best predictive error of dose changes based on bladder volume variations using deep learning exceeded 10%. Conclusion: 1. There is no significant difference in bladder volume and dose variations between the two IGRT strategies. 2. Bladder dose changes cannot be accurately predicted solely based on volume variations, emphasizing the need for separate dose verification to assess whether radiotherapy plans require adjustment. 3. Although bladder volume and dose exhibit some degree of inter-fraction variability, the majority of fractions show changes within 20%, indicating good consistency during IGRT for prostate cancer. However, special attention and appropriate adjustments are required for fractions with significant bladder volume or mean dose variations to ensure the precision and safety of radiotherapy. References: 1.Chen Z, Yang Z, Wang J, et al. Dosimetric impact of different bladder and rectum filling during prostate cancer radiotherapy[J]. Radiation Oncology, 2016, 11: 1-8. 2.Ma X, Chen X, Wang Y, et al. Personalized modeling to improve pseudo – computed tomography images for magnetic resonance imaging – guided adaptive radiation therapy[J]. International Journal of Radiation Oncology* Biology* Physics, 2022, 113(4): 885-892. 3.Song Y, Hu J, Wu Q, et al. Automatic delineation of the clinical target volume and organs at risk by deep learning for rectal cancer postoperative radiotherapy[J]. Radiotherapy and Oncology, 2020, 145: 186-192. Keywords: prostate cancer,IGRT,Bladder variation

3770

Poster Discussion 5 year outcomes of a randomized feasibility study comparing SABR with or without nodal irradiation in high risk prostate cancer (SPORT – NCT03253978) Nicola Hill 1 , Orla A Houlihan 2 , Ciaran Fairmichael 2 , Ciara A Lyons 2 , Conor K McGarry 3 , Darren Mitchell 1 , Aidan Cole 2 , Denise Irvine 3 , Michael Hanna 1 , Alan R Hounsell 3 , Joe M O'Sullivan 2 , Suneil Jain 2 1 1. Department of Clinical Oncology, Northern Ireland Cancer Centre, Belfast Health and Social Care Trust, Belfast, United Kingdom. 2 2. Patrick G. Johnston Centre for Cancer Research, Queen's University Belfast, Belfast, United Kingdom. 3 3. Department of Radiotherapy Medical Physics, Northern Ireland Cancer Centre, Belfast Health and Social Care Trust, Belfast, United Kingdom