Leadership • Cross-Collaboration • Winning Practices

VOLUME 19 • ISSUE 5 • 2019 • € 22 ISSN = 1377-7629 Silver Tsunami • TURNING THE SILVER TSUNAMI INTO A SILVER LINING, A. LOURENÇO • MANAGING THE WHOLE HEALTH OF THE AGEING POPULATION, C. BUCKLEY • OPPORTUNITIES AND RISKS OF DIGITAL HEALTH: OLDER PEOPLE’S PERSPECTIVE, E. HUCHET • ANTI-AGEING THERAPIES: FROM BASIC SCIENCE TO HUMAN APPLICATION, M. ABDELLATIF & S. SEDEJ • AGEISM IN HEALTHCARE: WHY IT HAS TO STOP, L. AYALON • SECRETS OF LONGEVITY - THE IKARIA STUDY, P. PIETRI • FRAILSAFE SYSTEM: AN INNOVATIVE APPROACH ON FRAILTY, S. MOZA ET AL.

SINGAPORE: TRANSFORM- INNOVATION AND INSPIRATION THE SEX AND GENDER INFLUENCE ATIVE SHIFTS IN HEALTHCARE FOR HEALTHCARE - HOW TO ON HYPERTENSION, S. SHAH MANAGEMENT, E. F. SOH CHANGE FERTILITY CARE WITH PARESH ET AL. VALUE-BASED HEALTHCARE, M. EUSOBI 2019, G. FORRAI CURFS HOW THE BRAIN WORKS: LOOKING INSIDE TO TARGET TREATMENTS, VENDOR-DRIVEN STANDARDS FOR INNOVATIVE HEALTHCARE STRA- S. MULDOON INTEROPERABILITY, D. HANCOCK TEGIES, P. FACON NEW MANAGEMENT PATHWAYS IN BREXIT: WREAKING HAVOC IN GAME-CHANGING MEETING OF CARDIOVASCULAR RISK FACTORS, HEALTHCARE?, M. DAYAN ET AL. : RADIOLOGY AND IMAGING R. VIDAL PEREZ INFORMATICS, E. RANSCHAERT LEADING CHANGE AS A PHYSICIAN, TACKLING THE FIVE ESSENTIAL X. CORBELLA & E. O' SULLIVAN IMAGE INFORMATION DELIVERY LEVERS OF THEATRE EFFICIENCY, IN THE AI ERA: TWO LIKELY D. THORPE

©For personal and private use only. Reproduction must be permitted by the copyright holder. Email to [email protected]. must be permitted use only. Reproduction and private ©For personal SCENARIOS, S. R. BAKER. WINNING PRACTICES

Erik Ranschaert Radiologist The Netherlands Cancer Game-Changing Meeting Institute - Antoni van Leeuwenhoek Amsterdam, The Netherlands President of EuSoMII of Minds: Radiology and Vienna, Austria

[email protected]

erikranschaert.com Imaging Informatics

eusomii.org

@eranrad A visionary radiologist explains how the European Society of Medical Imaging Informatics is bringing imaging and informatics together for best practices and patient outcomes.

What is the mission of the European setting is totally different from the clin- tainly facilitate the introduction of AI into Society of Medical Imaging Infor- ical work environment. radiological practice. That way they will matics (EuSoMII)? EuSoMII brings radiology and in- be able to develop products which are Today, everything is related to the devel- formatics professionals together to optimally adapted to the local needs. opment of Artificial (AI) solu- exchange knowledge and expertise and tions. This is a really hot topic and stake- to join the forces of these specialties What AI training routes is EuSoMII holders are looking for the correct way for better healthcare. Uniquely, we don’t offering? to implement this technology. There are just have radiologists among our mem- Radiology is split up into subspecial- many possibilities with the application of bers but other experts in the field such ties based on different body areas. We AI in radiology. On the one hand, there as clinical physicists and computer could say that imaging informatics has is a huge volume of images and data scientists. also become a subspecialty. In this era available and, on the other hand, there We also give members the chance of AI, there are (and will be) more and are a great number of AI developments. to increase their networks beyond the more tools we can use to analyse and However, data is not being shared with clinical and informatics fields. evaluate data. When you start using this, AI developers. This is the conflict be- you have to know what is happening with tween research and commerce too: How is EuSoMII bringing healthcare the data and how it is being processed. data is currently mainly being used for and technology stakeholders to- You really need some basic knowledge – research in the hospital/institutions and gether to improve cooperation for as much as you need knowledge about data needed for commercial products better radiology work processes? an MRI machine that you are using to is much more difficult to obtain. A new Our main objective is to help radiologists examine the body. As much as every market is developing, including compa- become more acquainted with imaging radiologist should be able to detect nies specialised in curating this data. informatics in its broadest sense, in- the MRI “artefacts” that can occur by In healthcare, the majority of and most cluding AI, and to teach them how to use knowing the basic physics of the tech- important AI solutions are the models it to optimise their services. Experts in nique, they should be able to know being implemented in medical imaging. the field are providing training and in- where AI-tools can go wrong and what This makes sense as deep took sights. At our annual meeting, we invite the underlying is. For this reason off in image analysis. speakers from different disciplines and Charles Kahn, editor of the RSNA Journal The demand for medical imaging is hold scientific sessions to show what is on AI is asking the developers to produce increasing and the workload is rising, so being developed in research. We address “explainable systems” (explainable AI or we need solutions to facilitate a more the difficulties, pitfalls and, importantly XAI), making it possible to understand efficient workflow. In short, we need in this age of data protection, ethical and the rationale behind the results pro- technology and other tools that can as- legal issues. We also organise classes duced by these tools (Kahn 2019a). sist us in analysing images more accu- about how to use and even develop AI In our events, seminars and webinars, rately and quickly. solutions for clinical practice. It is essen- we’re focusing on the necessary skills There is a lot of interest from the tial that radiologists are informed about for radiologists to become more familiar computer scientist and software engi- what tools are available to make use with imaging informatics and show that neer side in AI, which is clearly visible in of their own data for creating custom- you don’t need to be a computer sci- the research arena, but they don’t really made AI-solutions, without the need entist to understand and work in this know how this can be implemented in to learn any software developer’s skills. area. We are also cooperating with the the real world of radiology. The research Because the ability to do that will cer- European Society of Radiology on the ©For personal and private use only. Reproduction must be permitted by the copyright holder. Email to [email protected]. must be permitted use only. Reproduction and private ©For personal 414 HealthManagement.org WINNING PRACTICES

from the bad. This company is looking at how they could use this software in the medical sector. Techniques similar to those they developed for agriculture and successfully implemented for selecting and categorising different plants, can be applied to analysis of medical images. They have already developed a tool to automatically segment kidney tumours on CT images of the organs for example. This way radiology can learn from tech- nologies that have already successfully been applied in other sectors.

How do you see the role of the radi- ologist changing in the future? Radiologists will have to become more data-savvy and gain more skills in using all available image and patient data. They will move in the direction of data analysts. Now, we are still merely looking at morphological findings. But we are not EuSoMII Annual Meeting 2018 Rotterdam - The Board doing enough with the other data that we have available to examine a lesion. curriculum content. Owing to EuSoMII meetings, so solid leadership skills will How can we get more information an efforts, Imaging Informatics has now be needed in supporting the integration abnormality than just the morphology? been included in the European training of all patient related digital data and in How can we see if a tumour is active or curriculum as a subspecialty. This means using this data so that, for each patient, inactive, malignant or non malignant? that we have to provide content to train we will be able to advise our colleagues Through optimal integration of imaging residents in becoming experts in im- concerning the best treatment path to informatics for analysing the data, our aging informatics. follow. diagnoses will be more accurate and our It won’t be necessary for every radi- treatment pathways personalised for the ologist to learn but, in each department, What could radiologists learn from patients. We will also have to familiarise there should be at least one Imaging other sectors or hospital specialties ourselves with the concept of patients Informatics expert, otherwise it will be about AI development and applica- able to use their own health data and missing the tools to go ahead and invest tion? other types of services directly offered in the future. First of all, other hospital specialties to patients for analysing those data, could learn from radiology. The more which means that there will be other What leadership skills do radiolo- standards you have, the easier it is to players active in the healthcare market gists increasingly need? use data. Like radiologists, dermatolo- that probably might compete with ra- Better-integrated informatics will lead gists and ophthalmologists also use a diologists, which by some could be to more targeted treatment, adhering to lot of images, but they don’t yet have perceived as a threat of the profession. the principle of personalised medicine. standards for organising, sharing and Therefore, radiologists will need to be This linking of patient genomic data and archiving. Radiologists have refined proactive and cut off the road of those imaging is known as radiogenomics, standardisation and can share their ex- competitors by actively engaging with which is the direction in which medicine, perience. the new technologies. including radiology, is going in long- If you look at what imaging could term. The more accurately we are able to learn from industry, I can give an ex- How will the field develop over the analyse the data being provided through ample of a Belgian company that is very next decade? imaging, the more accurately we will be active in image analysis in multiple sec- AI will have a serious impact on the able to see how a disease is evolving. tors. It works in agriculture, automotives medical profession and, along with other In this framework, radiologists are and smart city sectors. In farming, it has medical specialists, radiologists will have increasingly involved in multi-disciplinary robotised tools to sort the good produce to reevaluate their role. How will we inte- ©For personal and private use only. Reproduction must be permitted by the copyright holder. Email to [email protected]. must be permitted use only. Reproduction and private ©For personal Volume 19 • Issue 5 • 2019 415 WINNING PRACTICES

grate and how should we invest in these normal cases and offering them only the other societies such as the Society for new developments, both from a knowl- highly suspicious cases for validation. Imaging Informatics in Medicine (SIIM) edge and financial point of view? The Think of Moorfields Eye Hospital in and will also increase our importance demand for medical imaging is continu- London. They are collaborating with in assisting other subspecialty socie- ously growing and in several countries DeepMind, Google (Suleyman 2018) on ties. Alongside this, we are focusing on there is already a capacity problem for patients with diabetes. They are cur- our educational role and advisory role in performing and reading all examinations. rently developing an to analyse imaging informatics and all its related AI-tools are needed to assist radiologists these scans automatically because, aspects. For example, we have just con- in dealing with this increasing workload. each day, 1000 of these scans are tributed to a major international, multi- An essential task for AI-developers will being made, which makes it very diffi- society paper on the ethics of AI for radi- be to provide tools that can help radiolo- cult to leave this to the ophthalmologist ology, which will most likely be published gists in prioritising the reporting of those himself/herself. This is a huge number this autumn in several major radiological examinations in which life-threatening so many tasks will be automised to fa- journals (Kahn 2019b). abnormalities can be found, such as pul- cilitate better service to patients, reduce monary emboli for example. waiting lists and make diagnoses more There are not enough radiologists accurate. to do all of our image reading. In breast References cancer screening for instance, there are What can healthcare and AI stake- Kahn C (2019a) Explainable AI. RSNA. Available from always two reads. In the future, maybe holders expect from EuSoMII in the pubs.rsna.org/page/ai/blog/2019/8/explainable_ai#

the first read will be done by AI which will future? Kahn C (2019b) Do the Right Thing. RSNA. Available facilitate the radiologists in dealing with Our next objective will be to develop from pubs.rsna.org/doi/full/10.1148/ryai.2019194001 the large volumes of mammograms that cooperation on an international level. Suleyman M (2018) A major milestone for the treatment of eye disease. Deep . Available from deepmind. need to be dealt with by filtering out the We will cooperate more intensively with com/blog/article/moorfields-major-milestone.

Continuation from p.383

key success factors for real innovation Long-Term Thinking We need to believe it ourselves and in the hospital sector. I want to identify In comparison to Singapore, where, act upon it. Let me be clear that as a three factors, as follows, that need to according to Prof. Fidelis, health author- regulator we are ready to tackle all of be addressed with particular ities are thinking up to 100 years ahead, this together with you. Let’s innovate in the Belgian context. we are lucky if we can think a legisla- and build our future together. ture of maximum five years ahead. For Speed of Innovation innovation to be successful, we need We are struggling with the lengthy to give these changes time to happen, period needed to develop, make deci- however, we need to know where we sions on, and implement reforms in will be in 20 or more years. It is essen- healthcare policy. This is not only a tial that we learn to look as far forward Key Points responsibility of state structure or as possible. political decision-making, but also an • Healthcare has always faced issue of our healthcare sector system Mindset operational, financial and as a whole. Stakeholders and deci- Even if challenges are multiple and staffing challenges; the key is how to face them with an sion-makers are convinced that we complex, as Dr Stoffels said, ‘It can innovative mindset. need change, but when we try to oper- be done.' Western Europe and coun- ationalise, decide on and implement tries like Belgium have a great deal of • Policy and healthcare sector red tape is frustrating change, we encounter multiple hurdles well-developed policy structures and innovation to a degree that and these cause us to lose time. I am social security mechanisms. These are could abort attempts to not convinced that today we are moving clearly under pressure and this requires transform the industry. forward quickly and efficiently enough policy updates. But we do dispose of • We need to look a minimum to ensure that our healthcare system so many resources to face these chal- of 20 years ahead to innovate will be future-proof in time. lenges successfully. effectively now in healthcare. ©For personal and private use only. Reproduction must be permitted by the copyright holder. Email to [email protected]. must be permitted use only. Reproduction and private ©For personal 416 HealthManagement.org