ESTRO 2025 - Abstract Book

S2536

Physics - Autosegmentation

ESTRO 2025

Conclusion: The large proportion of inconsistent delineation suggests an atlas is necessary to provide uniformity and guidance in DFS delineation. A DFS specific autocontouring tool is ideally needed as available bone autocontour is unlikely to be adequate. For HNRMS patients, where DFS are still developing in majority of patients, delineation of these structures is essential to enable dose evaluation and better understanding of treatment impact.

Keywords: atlas, dentofacial structures, delineation References:

[1] A Davey, S Pan, A Bryce-Atkinson et al. The need for consensus on delineation and dose constraints of dentofacial structures in paediatric radiotherapy: outcomes of a SIOP Europe survey. Clinical and Translational Radiation Oncology 2023:43:100681 [2] S M Kelly, R Effeney, M N Gaze et al. QUARTET: A SIOP Europe project for quality and excellence in radiotherapy and imaging for children and adolescents with cancer Eur J Cancer 2022:172:209-220 [3] https://siope.eu/activities/joint-projects/quartet/ [4] J Chisholm, H Mandeville, M Adams et al. Frontline and Relapsed Rhabdomyosarcoma (FaR-RMS) Clinical Trial: A Report from the European Paediatric Soft Tissue Sarcoma Study Group (EpSSG) Cancers (Basel) 2024:16(5):998

3700

Digital Poster 3D DeepLab-based Automated GTV Segmentation in Head and Neck Cancer Using PET/CT Imaging Prosper Oyibo 1 , Keun Woo Kim 2 , Sarah Hargreaves 3 , Philip Wheeler 4 , Owain Woodley 4 , Thomas Rackley 5 , Mererid Evans 5,6 , Emiliano Spezi 1 1 School of Engineering, Cardiff University, Cardiff, United Kingdom. 2 School of Aerospace, Transport and Manufacturing, Cranfield University, Bedford, United Kingdom. 3 Bristol Cancer Institute, University Hospitals Bristol and Weston NHS Foundation Trust, Bristol, United Kingdom. 4 Department of Medical Physics, Velindre University NHS Trust, Cardiff, United Kingdom. 5 Department of Oncology, Velindre University NHS Trust, Cardiff, United Kingdom. 6 School of Medicine, Cardiff University, Cardiff, United Kingdom Purpose/Objective: Auto-segmentation of the Gross Tumour Volume (GTV) in Head and Neck Cancer (HNC) is a critical advancement in radiotherapy planning. Several PET automated segmentation (PET-AS) methods have been proposed to overcome the limitations of manual outlining. However, the low spatial resolution of PET and high uptake in normal tissues can obscure tumor activity. To improve accuracy and reproducibility of tumor delineation, we developed a method