ESTRO 2025 - Abstract Book

S2499

Physics - Autosegmentation

ESTRO 2025

Conclusion: This work demonstrates comparable clinical performance of 2D and 3D DL autocontouring models in the context of prostate RT with a training carbon footprint 6 times higher for the 3D model. Training costs and environmental impact should be considered in the clinical adoption of DL auto-contouring models.

Keywords: Autocontouring, Prostate, Radiotherapy

References: 1. Zhang Y, Liao Q, Ding L, Zhang J. Bridging 2D and 3D segmentation networks for computation-efficient volumetric medical image segmentation: An empirical study of 2.5D solutions. Computerized Medical Imaging and Graphics . 2022;99:102088. doi:10.1016/J.COMPMEDIMAG.2022.102088 2. Lannelongue L, Grealey J, Bateman A, Inouye M. Ten simple rules to make your computing more environmentally sustainable. PLoS Comput Biol . 2021;17(9):6-13. doi:10.1371/journal.pcbi.1009324 3. Isensee F, Jaeger PF, Kohl SAA, Petersen J, Maier-Hein KH. nnU-Net: a self-configuring method for deep learning based biomedical image segmentation. Nat Methods . 2021;18(2):203-211. doi:10.1038/s41592-020-01008-z

2958

Digital Poster DL-based segmentation of orodental structures supports assessment of radiation dose to teeth and mandible and maxilla alveolar and basal sub-volumes Laia Humbert-Vidan 1 , Austin H Castelo 2 , Renjie He 1 , Ruth Aponte-Wesson 3 , Andrew Hope 4,5 , Erin Watson 6 , Kristy K Brock 2 , Stephen Y Lai 7 , Clifton D Fuller 1 , Mohamed A Naser 1 , Amy C Moreno 1 1 Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, USA. 2 Imaging Physics, The University of Texas MD Anderson Cancer Center, Houston, USA. 3 Oral Oncology, The University of Texas MD Anderson Cancer Center, Houston, USA. 4 Radiation Oncology, University of Toronto, Toronto, Canada. 5 Radiation