S1528
Interdisciplinary - Quality assurance and risk management
ESTRO 2026
cancer patients. The platform allows continuous monitoring of clinician edits promising to function as a useful QA tool. References: [1] S. G. Mikalsen et al. “Extensive clinical testing of deep learning segmentation models for thorax and breast cancer radiotherapy planning”. In: Acta Oncologica 62.10 (2023), pp. 1184–1193.[2] S. Jodogne. “The Orthanc Ecosystem for Medical Imaging”. In: Journal of Digital Imaging 31.3 (June 2018), pp. 341– 352. issn: 1618-727X. doi: 10.1007/s10278-018-0082- y.[3] M. J. Cardoso et al. MONAI: An open-source framework for deep learning in healthcare. 2022. arXiv: 2211.02701 [cs.LG].[4] Grafana. https://grafana.com/.[5] G. Palazzo et al. “Real-world validation of Artificial Intelligence-based Computed Tomography auto-contouring for prostate cancer radiotherapy planning”. In: Physics and Imaging in Radiation Oncology 28 (2023), p. 100501. Keywords: Autocontouring, Prostate Cancer, Imaging
Proffered Paper 3192
Designing a Regional Clinical Audit System in Radiotherapy: Tools and Infrastructure from the CAT-ClinART Project Carles Muñoz-Montplet 1,2 , Xavier Maldonado 3 , Cristian Candela-Juan 4 , Antonio Herreros 4 , Jaime Pérez-Alija 5 , Diego Jurado-Bruggeman 1,2 , Victor Hernandez 6 , Gemma Sancho-Pardo 7 , Núria Jornet 5 1 Medical Physics and Radiation Protection Department, Institut Català d'Oncologia, Girona, Spain. 2 Radiation Oncology and Medical Physics of Girona Group, Girona Biomedical Research Institute (IDIBGI), Girona, Spain. 3 Radiation Oncology Department, Hospital Universitari Vall d’Hebron, Barcelona, Spain. 4 Radiation Oncology Department, Hospital Clínic de Barcelona, Barcelona, Spain. 5 Servei de Radiofisica i Radioprotecció, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain. 6 Department of Medical Physics, IISPV, Hospital Universitari Sant Joan de Reus, Tarragona, Spain. 7 Radiation Oncology Department, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain
Figure 2 presents the dashboard illustrating OARs as example and associated metrics. Using well- established, open-source solutions for both DICOM operations (storage and query/retrieve) and web technologies (database and user interface), the platform developed a scalable and extensible system. Additional structures and other metrics can be seamlessly integrated, and the entire platform can be readily shared across institutions.The DSC of rectum, evaluated over time, showed a mean value of 0.94 ± 0.05, for bladder DSC = 0.96 ± 0.11; consistent with the validation study. Editing was null or minor for most cases: major editing (DSC < 0.80) was registered in 4 and 1 patients for bladder and rectum respectively. No time trends were seen. Different attitudes between clinicians were evident, mainly for femoral heads. Conclusion: This study demonstrated the feasibility of integrating a monitoring tool for AI-based autocontouring within the clinical workflow. The exploratory comparison between clinician-edited and auto-generated contours aligned with internal model validation for prostate
Made with FlippingBook - Share PDF online