ESTRO 2021 Abstract Book

S271

ESTRO 2021

Abstract Text Ignoranti quem portum petat, nullus suus ventus est. If a man does not know to what port he is steering, no wind is favourable to him. Seneca. Epistolae, LXXI., 3.

There a several metrics that are commonly used for the assessment of contouring, such as the Dice Similarity Coefficient or Hausdorff distance. Their ubiquitous presence in the research perpetuates their use and acceptance as the measurements to make. But are these measures the most suitable? That surely depends on the port we are sailing for! In this talk, we will look at the common approaches to assessing contouring similarity. However, our consideration will framed by the purpose of the assessment: specifically how this relates to clinical use of contouring. What is the impact on the clinical workflow of generating the measurement? What do we want to know about the contours? Although the focus will be primarily on quantitative measures, the use of qualitative and dosimetric assessments will also be considered. SP-0367 Clinical validation of automatic segmentation C. Robert 1 , A. Gasnier 1 , P. Blanchard 1 , S. Rivera 1 , A. Munoz 2 , V. Grégoire 2 , E. Deutsch 1 1 Gustave Roussy Cancer Campus, Department of radiation oncology, Villejuif, France; 2 Centre Léon Bérard, Department of radiation oncology, Lyon, France Abstract Text Deep learning (DL)-based auto-contouring solutions have recently been proposed as a convincing alternative to decrease the workload generated by the delineation of organs at risk and target volumes in radiotherapy treatment planning and improve inter-observer consistency. Following the 3rd European Society for Therapeutic Radiology and Oncology (ESTRO) Physics Workshop on "Implementation/Commissioning/Quality Assurance of Artificial Intelligence Techniques" in Budapest (2019), several recommendations were proposed regarding the implementation and quality assurance of artificial intelligence-based applications in radiotherapy. The first part of the presentation will summarize these recommendations for automatic contouring tools, in particular the methodology to be implemented to analyze the generalizability of the models and to evaluate their performance on local data, and the quality assurance programs to be performed for each patient or at a regular frequency. The second part of the presentation will illustrate the experiences of such a clinical implementation based on articles published in the literature. The local experience of the radiotherapy department of Gustave Roussy in France will conclude the presentation.

SP-0368 Quality assurance of image segmentations used for radiotherapy treatment planning J. Siebers USA

Abstract not available

Symposium: Working with infections/contagious pateitns - COVID-19

SP-0369 RTT practice during COVID-19 L. Capone Italy

Abstract not available

SP-0370 ESTRO RTTC guidance on COVID-19 M. Rossi 1 1 The Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands

Abstract Text Radiotherapy (RT) plays a key role in cancer management, and ideally, a course of RT should not be interrupted. Unforeseen interruptions to RT, leading to an increase in overall treatment time, can result in a higher risk of recurrences and poorer outcome. So with COVID-19 being classed as a pandemic in March 2020, it was important that RT services should be maintained and provided in a safe environment for both RTTs and patients. To this end, the European Society for Radiotherapy and Oncology, Radiation TherapisT Committee (ESTRO RTTC) published a guidance document and infographics at an early stage to aid RTTs in working safely as front line workers and keeping patients safe. Both these publications covered 4 domains: Patient care, RTT workflow, Remote working and RT practice and were translated into many European languages so that all RTTs in Europe would have access to them. Subsequently, a survey, initiated by ESTRO Radiation Oncology Safety and Quality Committee (ROSQC), was performed in May 2020 within the RTT community in Europe and America. The aim was to investigate the changes in RT practice in the 4 domains mentioned above, in the COVID-19 era and to recommend proactive measures to protect RT practice in future pandemics: 229 respondents from 27 countries completed the survey. The survey showed that within these 4 domains different countries had different approaches in dealing with COVID and RTTs were not always seen as front line workers despite their crucial role in the treatment of cancer patients. The survey also showed that in the early months lack of resources either personal protection, or financial issues dictated government guidelines. Most RTT departments followed government guidelines, but this was not always in line with acknowledging RTTs as front line workers. Systems and protocols needed to be put in place in a relatively short time to avoid spreading disease to staff and patients and the possible closure of a department because of no staff. The problems and