ESTRO 2025 - Abstract Book

S361

Brachytherapy - Physics

ESTRO 2025

Conclusion: AI-based ANR can achieve high agreement with CNR in many cases (within observer variation), showcasing potential for workflow enhancement, but expert supervision remains necessary to detect significant discrepancies.

Keywords: Automatic Needle Reconstruction, Cervical Cancer

References: Isensee, F., et al., “nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation,” Nat.Methods 18(2), 203–211 (2021). Jung, H., et al., “Deep-learning assisted automatic digitization of interstitial needles in 3D CT image based high dose rate brachytherapy of gynecological cancer,” PMB 64(21), 215003 (2019). Weishaupt, L. L., et al., “Approaching automated applicator digitization from a new angle: Using sagittal images to improve deep learning accuracy and robustness in high-dose-rate prostate brachytherapy,” Brachytherapy 21(4),520–531 (2022).