Trustworthy AI in Cancer Imaging Research

Ioanna Chouvarda (F), Electrical Engineer, PhD, is an Associate Professor of Medical Informatics and Biomedical Data Analysis, at the School of Medicine, Aristotle University of Thessaloniki. She has been involved in medical informatics research for over 25 years. She is particularly interested in a) biomedical data analysis and development of machine learning methods with biomedical images, biosignals and biological data, and b) digital health applications and services incorporating such models. She has worked in numerous national/European research projects in the field of digital health and biomedical informatics, has coordinated AUTH’s work in seven projects and has been an evaluator of numerous proposals and projects. She has developed a number of undergraduate and graduate courses in biomedical informatics/engineering, some with an interdisciplinary scope, and has supervised many graduate students. Sara Colantonio is a senior researcher at the Institute of Information Science and Technologies "Alessandro Faedo" of the National Research Council of Italy, based in Pisa and a member of the "Signals and Images" Laboratory, where she is coordinating the activities of a multidisciplinary team of researchers on the topics of artificial intelligence for health and wellbeing. Her research interests include artificial intelligence, machine learning, assistive technologies, personal informatics, quantified self, and trustworthy AI. She coordinated an FP7 project focused on the first sensorised smart mirror for cardiometabolic disease prevention, which granted her an award as one of the 40 Top Transformers in health in 2016. She is a member of the international multi-stakeholder group of experts "Artificial Intelligence for Health Imaging" (AI4HI), contributing to the definition of the FUTURE-AI guidelines on trustworthy AI in medical applications.Gianna Tsakou is a senior project manager, analyst and researcher of EC and nationally funded IT research projects with over 25 years of experience in both the public and private sector. She is currently working as a Senior Project Manager / Senior Analyst in the Research and Development Lab of Maggioli S.P.A., where she has the overview of the eHealth EC funded projects, notably those related to prevention, diagnosis and treatment of cancer using big data and AI. She has coordinated and/or participated in more than 30 R&D projects through which she has acquired expertise and know-how in a wide range of IT research areas among which: Artificial Intelligence (AI) in cancer imaging, trustworthy AI, health-related big data repositories, healthy ageing and independent living, eHealth services, accessible services for people with disabilities / eInclusion, multi-modal interfaces. She is a co-author in more than 30 scientific publications and one of the 4 co-initiators and coordinators of the "Artificial Intelligence for Health Imaging" (AI4HI) network.Dr. Guang Yang is an Associate Professor (Senior Lecturer) in the Bioengineering Department and Imperial-X at Imperial College London. He holds a UKRI Future Leaders Fellowship and serves as an Honorary Senior Lecturer in the School of Biomedical Engineering & Imaging Sciences at King's College London. He is an Associate Editor of IEEE Transactions and npj Digital Medicine. His research group is dedicated to developing novel and translational techniques for imaging and biomedical data analysis. The group’s focus encompasses research and development in data-driven fast imaging, data harmonization, data synthesis, federated learning, explainable AI, and AI in drug discovery. Currently, his work spans a wide range of clinical applications in ageing, cardiovascular disease, lung disease, and oncology.

• Section 1. Overall Considerations.- 1. Generating the FUTURE AI. describing the process for reaching consensus on the FUTURE-AI recommendations and how these contribute/relate to trustworthy AI (make some kind of correspondence to the trustworthy AI principles of the EC and others) Martijn Starmans, Richard Osuala, Oliver Díaz, Karim Lekadir, and contributors.- 2. The Clinical Viewpoint / Considerations for Clinical Impact of AI in Oncologic Imaging Luis Marti-Bonmati (clinical Ai4HI WG), and contributors from all AI4HI.- 3. Socio-ethical and legal implications of Trustworthy AI – the AI4HI ELSI Mónica Cano Abadía(BBMRI-ERIC, EuCanImage), Ricard Martínez (Primage and Chaimeleon) and Mario Aznar +ProCancerI legal colleague , and provisionally Magda Kogut (INCISIVE).- Section 2. Preparing for trustworthy AI: The Data and Metadata for quality, transparency and traceability.- 4. Data harmonization and challenges towards generation of repositories: sharing practices and approaches- ( Include Data de-identification / Include Data annotation and segmentation / compare commonalities and differences in the projects/ Data quality) Leonor Cerdá (Primage), Oliver Diaz( EUCANIMAGE), Guang Yang (Imperial, Chaimeleon), Ana Jimenez -Quibim /UNS/ Alexandra Kosvyra [AUTH] , Ch Kondylakis FORTH, provisionally co-authors from CERTH.- 5. Standardising Data and Metadata (this will include Data models/AI metadata / AI Passport /Transparency of Data, Models, and Decisions) Ch Kondylakis (FORTH), S Colantonio–(CNR) Gianna Tsakou (MAG) + Alexandra Kosvyra [AUTH] + provisionally inputs from ( Ticsalud/ED/ Medexprim/) Pedro Mallol (Chaimeleon).- 6. Generatic synthetic data in Cancer Research Yang (Imperial College)/ Leonor Cerdá, Richard Osuala , provisionally Karim Lekadir / Adrián Galiana (Primage).- Section 3. Implementing trustworthy AI: The Algorithms and DSS.- 7. Architectures and platforms for trustworthy AI: cloud technologies and federated approaches (this includes The privacy preserving methods / challenges with federated learning , Cloud technologies for supporting centralized trustworthy AI training ) Alberto Gutierrez (BSC) and Chrysostomos Symvoulidis (INCISIVE)/ Martijn Pieter Anton Starmans EUCANIMAGE / Ignacio Blanquer (CHAIMELEON ).- 8. AI robustness, generalizability and explainability Sara Colantonio, Alberto Gutierrez-Torre [BSC], And inputs from Nikos Papanikolaou. Ysroel Mirsky (Israel, Chaimeleon), Henry Woodruff (Maastrich, Chaimeleon), D Dominguez Herrera (Ticsalud) / D Fotopoulos (AUTH) / Manikis/KMarias (FORTH).- 9. AI Models in cancer diagnosis and prognosis Leonor Cerdá (Chaimeleon), D Filos and I Chouvarda (AUTH), Turukalo, Tatjana (UNS) and contributors from all projects (including ICCS fromINCISIVE project).- Section 4. Validating trustworthy AI: The Validation and User perspective.- 10. Doing Technical validation for real. Experiences from a multisite effort Inputs from the AI4HI WG survey work and relation to project work / AUTH and UNS can contribute the INCISIVE prevalidation method and efforts here (Olga Tsave/Chouvarda – AUTH) and (Tatjana Turukalo and UNS team), with contributors from all projects.- 11. Clinical Validation – (including material from previous AI4HI paper, User perspective/feedback and lessons learnt / experience difficulties from all projects) Luis Bonmati, Katrine Riklund , Shereen Nabhani-Gebara, Lithin Zacharias, Maciej Bobowicz,.- 12. Real-life deployment of AI services: practical implications (focusing on real-life deployment of AI services: practical implications, patents, fast-track for clinical usefulness, Towards certification) ( Ana Blanco, Ana Jimenez, Fuensanta Bellvis , Quibim) + legal partners from all teams on AI related requirements.