ESTRO 2025 - Abstract Book

S2463

Physics - Autosegmentation

ESTRO 2025

Figure 1: Metric comparison evaluated between manual-automatic, manual-edited and automatic-edited for CTV breast.

Conclusion: This study assessed the feasibility of introducing an automatic segmentation tool for breast planning into the clinical workflow. Results showed that the comparison between automatic and manual structures was worse than IOV values for CTV, requiring an editing longer than the one of manual segmentation. Instead, automatic contouring of heart, lungs and contralateral breast may robustly replace manual contouring.

Keywords: Breast-radiotherapy, Interobserver variability

References: [1] S. Mikalsen. “Extensive clinical testing of Deep Learning Segmentation models for thorax and breast cancer radiotherapy planning”. In: Acta Oncologica 10 (2023), pp. 1184–1193. doi: 10.1080/0284186X.2023.2270152.

[2] P. Buelens. “Clinical evaluation of a deep learning model for segmentation of target volumes in breast cancer radiotherapy”. In: Radiotherapy and Oncology 171 (2022), pp. 84–90. doi: 10.1016/ j.radonc.2022.04.015.

[3] S. Almberg. “Training, validation, and clinical implementation of a deep-learning segmentation model for radiotherapy of loco-regional breast cancer”. In: Radiotherapy and Oncology (2022), pp. 62–68. doi: 10.1016/j.radonc.2022.05.018.