Abstract Book

S117

ESTRO 37

uptake and its quantification by SUV will be discussed. Moreover, the need for validation of simplified uptake metrics and tumor delineations and its use as quantitative metabolic imaging biomarker will be [1] Boellaard R. (2009) Standards for PET image acquisition and quantitative data analysis. J.Nucl.Med. 50: 11S-20S [2] Boellaard R, O'Doherty MJ, Weber WA, et al. (2010) FDG PET and PET/CT: EANM procedure guidelines for tumour PET imaging: version 1.0. Eur J Nucl Med Mol Imaging. 37:181-200 [3] Boellaard R, Delgado-Bolton R, Oyen WJ, et al. (2015) European Association of Nuclear Medicine (EANM). FDG PET/CT: EANM procedure guidelines for tumour imaging: version 2.0. Eur J Nucl Med Mol Imaging 42:328- 54 [4] Hristova I, Boellaard R, Vogel W, et al. (2015) Retrospective quality control review of FDG scans in the imaging sub-study of PALETTE EORTC 62072/VEG110727: a randomized, double-blind, placebo-controlled phase III trial. Eur J Nucl Med Mol Imaging. 42:848-57 SP-0216 Image guided radiotherapy to improve treatment delivery accuracy, how far can we go? C. Dickie 1 1 Princess Margaret Cancer Centre, Radiation Medicine Department, Toronto, Canada Abstract text This lecture will focus on the accuracy of image guidance for radiotherapy. An overview of available IGRT technologies will be provided, including but not limited to ultrasound, Plain KV/MV, CBCT, SGRT, and MRI. The advantages and disadvantages of IGRT will be explored in the context of safety, quality and value added to radiation treatment. Evidence supporting the role of IGRT in the radiotherapy delivery process will also be considered, with the overarching theme of 'how precise can we possibly be' and 'how far can we actually go'. SP-0217 Big data: how to handle, interpret and analyse? A. Dekker 1 1 Maastricht Radiation Oncology MAASTRO, Knowledge Engineering, Maastricht, The Netherlands Abstract text Big data, artificial intelligence, machine learning and data science are new fields which are expected to have a major impact on day-to-day oncology practice. Big data based services such as automated contouring and planning, radiomics, decision support systems and literature mining are products already available to our community and these are expected to rapidly change the way we practice medicine. Gaining a basic understanding of Big Data and technologies based in it, including their strengths and weaknesses, is the overall aim of this teaching lecture. Specifically, the lecture addresses the following questions: - What is the rationale behind using Big Data for data driven medicine and what is the relation to evidence base addressed. References Teaching Lecture: IGRT, SGRT, IGART, improving treatment delivery accuracy, how far can we go? Teaching Lecture: Big data: how to handle, interpret and analyse?

medicine? - What is Big Data in (radiation) oncology? What types of data are there? How big is the data really? And where is it? - How do you deal with privacy protection? - What is the quality of Big Data? - How can you get access to Big Data? How does one learn from Big Data? - How do you critically appraise and apply Big Data results into daily practice? The teaching lecture is suitable for a wide audience, and targeted to practicing radiation oncologists, medical physicists and RTTs who wish to get an overview on the topic.

Symposium: European Particle Therapy Network (EPTN)

SP-0218 EPTN WP1: Clinical trial designs to assess the benefit of protons H. Langendijk 1 1 UMCG University Medical Center Groningen, Department of Radiation Oncology, Groningen, The Netherlands

Abstract text Background

Randomized controlled trials (RCT’s) are considered the gold standard of evidence-based medicine. This is certainly true if new radiation technologies like protons are introduced to improve efficacy in terms of local tumor control and/or overall survival. However, most new radiation technologies are clinically introduced to further reduce the unintended dose to the normal tissues primarily aiming at reduction of radiation-induced side effects. There is growing recognition that for such purpose, RCT’s are probably not the most ideal study design, given major variability in performance between centers, rapid technological developments and variability in the level of expected clinical benefit for individual patients. Therefore, alternative evidence-based methods are needed and available. Model-based approach The model-based approach is an alternative evidence- based methodology developed to select patients for proton therapy and to clinically validate the added value of new radiation-induced technologies aiming at reduction of radiation-induced side effects. Model-based selection Model-based selection consists of 3 steps. Step 1 includes the selection of high quality NTCP-models. Step 2 includes a planning comparative study for each individual patient comparing photon with proton plans to assess the difference in dose volume histogram (DVH) parameters derived from the NTCP-model (DDose). Step 3 includes the integration of the results of the individual planning comparative study into the NTCP-model selected to translate DDose to DNTCP. The level of DNTCP can then be used for selection of patients. Model-based validation The principle of model-based validation is to test the hypothesis that the observed rate of radiation-induced side effects obtained with the new technology (i.e. protons) is lower than the average NTCP expected from the old technology (i.e. photons). To this end, patients selected for protons are included in a prospective observational cohort study, in which the performance of the photon-based NTCP-models are continuously tested on model performance using the closed testing procedure. The closed testing procedure is a method to assess whether NTCP-model adjustment is required and to what level, varying from no adjustment, recalibration- in-the-large, recalibration, model refit or model revision.