ESTRO 2025 - Abstract Book

S3423

Physics - Machine learning models and clinical applications

ESTRO 2025

3429

Proffered Paper Multicenter prospective risk analysis of the fully automated radiotherapy treatment chain Geert De Kerf 1,2 , Ana Barragan 3 , Charlotte Brouwer 4 , Pietro Pisciotta 5 , Marie-Claude Biston 6 , Marco Fussella 7 , Geoffroy Herbin 8 , Esther Kneepkens 9 , Livia Marrazzo 10 , Joshua Mason 11 , Camila Panduro Nielsen 12,13 , Koen Snijders 1 , Stephanie Tanadini-Lang 14 , Aude Vaandering 15 , Tomas M Janssen 16 1 Radiation Oncology, Iridium Netwerk, Wilrijk, Belgium. 2 Faculty of Medicine and Health Sciences, University of Antwerp, Antwerp, Belgium. 3 – Institut de Recherche Expérimentale et Clinique - Molecular Imaging Radiotherapy and Oncology, UCLouvain, Brussels, Belgium. 4 Radiation Oncology, University Medical Center Groningen,, Groningen, Netherlands. 5 Radiation Oncology, University Medical Center Groningen, Groningen, Netherlands. 6 Radiation Oncology, Centre Léon Bérard, Lyon, France. 7 Radiation Oncology, Abano Terme Hospital, Padua, Italy. 8 Research and Development, Ion Beam Applications, Louvain-la-Neuve, Belgium. 9 Radiation Oncology, Maastro Netwerk, Maastricht, Netherlands. 10 Experimental and Clinical Biomedical Sciences, University of Florence, Florence, Italy. 11 Imperial College Healthcare, NHS Trust, London, United Kingdom. 12 Clinical Research, University of Southern Denmark, Odense, Denmark. 13 Radiation Oncology, Odense University Hospital, Odense, Denmark. 14 Radiation Oncology, University Hospital Zurich, Zurich, Switzerland. 15 Radiation Oncology, Cliniques Universitaires ST Luc, Brussels, Belgium. 16 Radiation Oncology, The Netherlands Cancer Institute, Amsterdam, Netherlands Purpose/Objective: Participants of the ESTRO 2023 physics workshop “AI for the fully automated radiotherapy treatment chain” recognized that a fully automated radiotherapy treatment planning workflow (FAW) is technically possible, but clinical implementation is limited [1,2]. A multicenter prospective risk analysis for the FAW was initiated, aiming to support centers in managing FAW-related risks and to identify steps requiring additional tools or improvements. Material/Methods: Eight European radiotherapy centers conducted a failure mode and effect analysis (FMEA) on a hypothetical FAW, starting from a CT image and ending by a manual review (Figure 1).

Each center was asked to evaluate severity (S), occurrence (O) and detectability (D) for provided failure modes, identified during the Physics workshop, to obtain a risk score S*O*D [3]. In each center, a multidisciplinary team conducted the risk analysis. Data curation of the collected scores consisted of grouping potential causes and effects in respectively 5 and 10 major categories. To minimize variation in scoring risk factors between different centers, O, S and D values were normalized per center and afterwards the overall risk score was recalculated for the quantitative analysis. Moreover, centers could add free text for a qualitative assessment.