ESTRO 2025 - Abstract Book

S4401

RTT - Treatment planning, OAR and target definitions

ESTRO 2025

Fig 1: Interobserver variability in the manual correction of cardiac substructures by five radiation oncologists on a contrast-enhanced CT scan.

Conclusion: This automatic contouring algorithm demonstrated high accuracy in delineating cardiac substructures on contrast enhanced CT scans. Structures identified as requiring closer review before dose extraction were primarily segments of the coronary arteries, including the Right Coronary Artery, middle-distal segment, proximal and middle-distal Circumflex Artery.

Keywords: Deep learning, contouring, radiotherapy planning

References: Vaugier L. et al. How to contour the different heart subregions for future deep-learning modeling of the heart: A practical pictorial proposal for radiation oncologists. Clin Transl Radiat Oncol . 2024; 45:100718.

4204

Digital Poster Validation Of An In-House AI-Segmentation Algorithm For The Delineation of Head and Neck Organs At Risk Matthew Sciberras 1 , Gašper Podobnik 2 , Tomaž Vrtovec 2 , Susan Mercieca 1 1 Radiography, Faculty of Health Sciences, University of Malta, Msida, Malta. 2 Faculty of Electrical Engineering, University of Ljubljana, Ljubljana, Malta Purpose/Objective: Segmentation of head and neck (H&N) organs-at-risk (OARs) for radiotherapy planning is time-consuming. This study evaluated the accuracy of an in-house developed artificial intelligence (AI) algorithm for segmenting the parotid glands, submandibular glands, mandible, brainstem, optic nerves, and optic chiasm. Material/Methods: Ten computed tomography images of H&N with the OARs contoured by experts were identified from the University of Ljubljana segmentation challenge (https://han-seg2023.grand-challenge.org/). The accuracy of the AI