V O L 2 9 I S S U E 1 / 1 9 F e b / M a r 2 0 1 9 T H E E U R O P E A N F O R U M F O R T H O S E I N T H E B U S I N E S S O F M A K I N G H E A L T H C A R E W O R K5-24 ECR SPECIAL reports and more RADIOLOGY • Artificial intelligence: • Mass Spec detects illicit drug presence LAB & DIGITAL PATHOLOGY 25-28 the time has come • Man and machine: today’s centaur • Mapping the inflammatory landscape • AI for digital pathology • Radiomics on tap in 5-10 years The end of medicine as we know it The physician is no longer the lone master of diagnosis Physician, chemist and pharmacist Professor Harald Schmidt chairs the Department of Pharmacology and Personalised Medicine, at the Faculty for Health, Medicine and Life Sciences, University of Maastricht. As advanced investigator of the European Research Council (ERC) he also leads various research programmes, among them a Proof-of-Concept study on strokes; the COST Action OpenMultiMed on system medicine and the Horizon 2020 programme REPO-TRIAL. Schmidt was co-founder and managing director of Vasopharm GmbH, is co-publisher of the journal Systems Medicine and has written more than 200 refereed international publications, summary articles and books. Announcement: Professor Harald Schmidt (Maastricht) will discuss ‘The end of medicine as we know it’ at the ETIM 2019 - Artificial Intelligence and Smart Hospital Summit Saturday 23 February 2019. Time: 3:55-4:30 pm puter does not administer therapy; AI proposes a diagnosis with a probability of x and the therapy appropriate to it. The decision as to which way to go is reserved for the medical team, the patients and their priorities.’ www.healthcare-in-europe.com 1-3 4 CONTENTS NEWS & RESEARCH ONCOLOGY RADIOLOGY – PLUS ECR 5-24 NEWS AND VIEWS LABORATORY 25 DIGITAL PATHOLOGY 26-28 researchers and Currently many experts assume that the next great socio-economic revolution will include a completely new defini- tion of health and how we define illnesses and therapies. According to Professor Harald Schmidt, head of the Department of Pharmacology and Personalised Medicine, Faculty for Health, Medicine and Life Sciences at the University of Maastricht, ‘Our health system today can no longer be sustained in its existing form. It has become too expensive and too ineffective.’ The end of medicine, as we know it? Asked for a fuller explanation of his prediction, he pointed to a fun- damental problem. ‘Our definition of illness is based on organ localisation or individual symptoms. All medi- cine, as we teach it, in the way we train specialists, the way our clinics are structured, is divided according to organs. There is a specialist physi- cian and clinic for every organ. ‘However, illnesses do not func- tion that way, rather we experience disruptions in the signal systems in our cells – put more precisely, in certain hotspots that are disturbed and cause symptoms. In most cases these signal disruptions occur in more than one organ and produce different symptoms. Thanks to the organ-based division, we do not see that though. And thus there are hardly any systematic or holistic approaches to medicine. ‘The pharmaceutical industry is also contributing to the end of medicine as we know it, doomed www.healthcare-in-europe.com Inefficiency of current pharmaceutical treatments: the green patients represent those for whom a medicine works; although the reds are prescribed that same medication it delivers no benefit adapted from Scannell et al. (2012) Nat Rev Drug Discov 11(3):191-200 in its current form. In the list of the world’s ten largest companies there is not a single pharmaceutical com- pany any more. That is somehow logical because why should we be forever discovering new medicines? ‘One reason why, after 100 years of pharmaceutical industry, we are still researching for new medicines is the lack of precise illness defini- tions and the chronic treatment of symptoms instead of healing. To date we know the molecular causes of very few illnesses and therefore cannot treat them.’ Medical education must change ‘Medical education will have to change along with hospital struc- tures. Physicians will have to work in interdisciplinary teams more often, together with bioinformatics specialists and system physicians. Actually the days when the medical specialist treats a patient based on his personal current knowledge are now already over. ‘A research phase will precede these changes. But, it’s clear to see that such a medicine will lead to significantly more interdiscipli- nary hospital structures. Thus, the first clinics have already introduced immune disease boards. In these boards, illnesses such as psoriasis, asthma, colitis or rheumatoid arthri- tis are treated by a team, work- ing like the familiar tumour boards for oncological issues. Even experts agree that these disease designa- tions are obsolete. These are essen- tially immune system diseases that have to be defined molecularly and not by clinical morphological means. ‘We also have to bid farewell to classical pathology, currently exhausted by histology and the detailed description of cell changes. More than anything we need a molecular pathology that leads to molecular disease definitions and thus permits precise diagnosis and therapy. To attain this, we have to work more in terms of systems and more holistically. ‘Once we have penetrated deeply enough to identify the disease mech- anisms, then we can apply a pre- cisely suited medication – possibly one that has already been approved. ‘In the not too distant future there will be at least one appropriate medication for all the disease mech- anisms so defined, the number of which will be between 160 and 200. Thus a research and development pharmaceutical industry is obsolete; only manufacturers and distribu- tors will still be needed. Only those pharmaceutical enterprises that heal rather than treat will survive. ‘The first steps have been taken in this direction. In 2018, the USA’s FDA approved tumour medications that are no longer approved to treat an organ tumour – but for an oncological mechanism, which means they are approved for the specific mechanism, regardless of where the tumour is found. That is a small turning point for medication therapy.’ The AI or big data impact on health systems ‘Despite necessary changes, initially, the treatment cycle will naturally stay the same: as a rule the patient with organ-based symptoms will look for a physician, only his diagnosis will be strongly supported by artificial intelligence. Perhaps AI will propose to the physician additional measure- ment parameters, digital imaging or genetic examinations and thus determine a molecular illness diag- nosis. Then just like today it will be the task of the medical team to support the patient in choosing the optimum therapy for him. ‘The physician is no longer the lone master of diagnosis – he can- not be that any more. The net- worked data analysis necessary to analyse big data will be possible everywhere. Nonetheless there will still be precise ad hoc diagnoses and manual therapies, such as the surgical specialties. However, there will be big changes in the field of chronic illnesses.’ AI will not make a therapy decision ‘Currently the most common cause of death is physician’s error – we already have the problem. The com-

N E W S & M A NAG E M E N T 3 Spatial cognition in aging and neuro-degeneration When the compass fails Where are we coming from? Where are we going? Where are we right now? Our sense of spatial orientation - a complex interaction of various cells in the brain - gives us answers to these questions. In November 2018, around 70 experts in Magdeburg discussed changes in this naviga- tion system induced by age and illness at the ‘Interdisciplinary Symposium on Spatial Cognition in Aging and Neurodegeneration’ (iSCAN). Professor Thomas Wolbers, one of the organisers, spoke with us about the problems of diagnosing orientation disturbances and how the measurement of our sense of space could help in the fight against neurodegenerative illnesses such as Alzheimer’s. much as psychiatric illness such as schizophrenia. Using VR for standardised orientation tests While the activities of the sense of orientation can be shown using fMRI, the diagnosis of the distur- bances is still difficult due to a lack of standardised tests. ‘That ought to change in the coming years,’ the neuroscientist hopes. ‘There are sev- eral groups working on the develop- ment of tests by which it is possible to evaluate the seriousness of orien- tation disturbances.’ One of the key technologies here is virtual reality (VR): If the person being tested is given the task with 3-D visors to find his way in a virtual space, then much data is generated that can be evaluated that allows measurement of the kind and degree of dis- turbance. Possibly VR may even be used for in cases of orienta- tion disturbanc- es. But there is still a long way before this is possible, W o l b e r s believes. therapy to go D u r i n g the recent i S C A N Symposium, early recognition of cognitive dete- rioration processes was also debated. A uniform standard to measure ori- strongly Professor Thomas Wolbers heads a working group at the German Centre for Neurodegenerative Illnesses (Deutsches Zentrum für Neurodegenerative Erkrankungen (DZNE)) in Magdeburg and is among the organisers of the ‘Interdisciplinary Symposium on Spatial Cognition in Aging and Neurodegeneration’ (iSCAN). A specialist in age-related and illness-induced changes in cognition, his research investigates neurodegenerative processes, the development of standardised neuropsychological diagnostics and training methods to avoid neuronal deficit as well as the restoration of cognitive functions. first diagnosis entation disturbances could also serve as a significant prognosis marker. ‘Alzheimer’s has a degenera- tive effect’, Wolbers noted. ‘This means that the illness has already been progressing in many patients for 10 to 15 years before the is made. Currently there is still no therapy. But, when a treat- ment is devel- oped, it is of e n o r m o u s impor tance that the signs of the illness are recog- nised as early as possible. Disturbances in the sense of ori- entation could be such an indicator.’ Source: Shutterstock/Nizwa Design Report: Wolfgang Behrends ‘In our brain there are several cell types that code spatial information,’ explains Wolbers. The so-called place cells in the hippocampus and grid cells in the entorhinal cortex process the current position in the space – in principle similar to the GPS in a navigation system. Other cells function like a compass, send- ing signals to the head as to which direction it is oriented at any given time. This collective spatial knowl- edge is used by the brain to coor- dinate motoric impulses and thus to steer the movement of the body from A to B. Just like a compass that no longer points North reliably after several shocks, even this orientation centre in the head can malfunction: ‘Due to brain damage in the regions where these types of cells are located, patients can develop very specif- ic orientation disturbances.’ Using functional magnetic resonance imaging (fMRI), these disturbances can be rendered visible: Changes in the blood’s oxygen content exhibit the activity of certain brain areas in the form of voxels (spatial pix- els). ‘Thus we can recognise signals transmitted by grid cells when tra- versing a space,’ Wolbers explained. If there is a disturbance in the sense of orientation, then these signals appear as distorted or temporally unstable. ‘That shows that the cells are no longer able to precisely com- pute the position in space.’ The vicious circle of deterio- ration The most important disturbance fac- tor is aging, Wolbers pointed out. ‘When we age, different neuronal and causality in data patterns,’ Schreiweis observed. Current scepti- cism towards artificial intelligence is understandable, but unfound- ed, said Walia: ‘When MR imaging was first introduced, most experts were also sceptical – but now you couldn’t imagine medicine without it. Medics who deny the potential of artificial intelligence will become obsolete one day.’ www.healthcare-in-europe.com In the entorhinal cortex (red) the brain processes the current position in space. This region is particularly vulnerable to age-related and illness-induced degeneration, which can lead to orientation disturbances. deterioration processes occur in the brain. Some of the regions, like the entorhinal cortex, are particu- larly vulnerable to such deteriora- tion. Thus the spatial orientation capacity declines successively with age.’ Usually the disturbances occur first in unfamiliar environments, for example on vacation; later orien- tation becomes increasingly diffi- cult even in familiar places. ‘This leads many senior citizens to adopt an avoidance attitude, so they are almost unable to move beyond their own homes. This again promotes cognitive deterioration – a vicious circle.’ Illnesses can further accelerate this disorientation. ‘The best known example is certainly Alzheimer’s dis- ease. Here abnormal tau proteins are deposited in the brain that can restrict the functionality of the nerve cells – and thus also cogni- tive performance. In some six in 10 Alzheimer’s patients this illness also has an impact on orientation – they get lost frequently, don’t find their way home without help anymore.’ Depending on which brain region is affected, even a stroke can restrict the compass in the head, just as They have no bread? Give them artificial intelligence! The potential of artificial intelli- gence is vast, and not only for medi- cine, Walia confirmed. But before the technology can be introduced comprehensively some very differ- ent issues must be addressed. ‘There is no point in convincing someone of the advantages of artificial intel- ligence if that person does not have enough food to live on.’ Education is a decisive prerequisite required to enable people to handle new technology. An important part of technological progress is to design systems in such a way that they can be used with ease and intuitively, emphasised Nurkka. Anonymous health data – an illusion? People need to understand the importance of personal medical data, Heinemann said. ‘These days, companies think of health data as a resource – but this view hinders technological progress. We need free access to anonymised health data to train algorithms.’ But, this is where the problem lies, Schreiweis countered, because real data anonymity is an illusion: ‘When sufficient data sets are availa- ble, algorithms can match individual data to individuals.’ Companies such as Google sup- port hospitals with generous dona- tions, Walia noted. This facilitates important progress, from which everyone will benefit ultimately. However, in return their money buys them access to research – and health data – which they use to their own advantage. Adam put it in a questioning nutshell: ‘How do we facilitate the secure handling of personal health data without limiting access to data sources?’ Hope for a better future To really utilise the enormous potential of AI in medicine, we need a system which does not view the data as a commodity, Heinemann added. ‘Then everyone would have access to the treatment they need, independent of their financial sit- uation.’ Nurkka and Adam were equally confident – assuming that data was to be liberated from its proprietary silos in companies or hospitals and made available widely. ‘We just need to connect the data points,’ Adam concluded. ‘Then we have all the information we need for better healthcare.’ BIOTRONIK Home Monitoring® Simplicity of Care. Anytime. Anywhere. BIOTRONIK´s award-winning remote patient management system is the most user-friendly and clinically actionable cardiac remote monitoring solution available today. www.biotronik.com

@ ECR W W W. H E A LT H C A R E - I N - E U R O P E . C O M 2019 S P E C I A L I S S U E F O R T H E E U R O P E A N C O N G R E S S O F R A D I O L O G Y V I E N N A A U S T R I A F E B 2 7 – M A R 0 3 2 0 1 9 Man and machine: today’s centaur Artificial intelligence continues to drive radiologists’ discussions. Among them, Associate Professor Georg Langs, head of the Computational Imaging Research Lab (CIR) at the University Clinic for Radiology and Nuclear Medicine at the Medical University of Vienna, believes: ‘The evaluation of patterns in data from imaging examinations and clinical information about patients using machine learning will change fundamentally our understanding of illnesses and their treatment as well estimation of their course.’ Report: Michael Krassnitzer e r e h p s i m e H t f e l The Austrian computer scientist sees two major applications for machine learning – the form of AI by which a computer program learns from a mass of examples. First is the automated recognition of already familiar patterns, markers or sig- natures in image files that are diagnostically relevant, that also help to say something about the future course of an illness or, for example, how a certain patient will respond to a treatment. This also includes even the search for small tumours or metas- tases. The second applica- tion – and in Langs’ view perhaps even more prom- ising – is the discovery of new patterns, markers or sig- natures of diagnostic relevance that have yet to be identified. ‘Machine learning delivers really good results where we have come to a dead end using conventional markers,’ Langs emphasises. A good example is the diagnosis of idiopath- ic pulmonary fibrosis (IPF), a difficult to diagnose and rare lung disease. Here Langs‘ research group, in close cooperation with the lung fibrosis specialist Professor Helmut Prosch at the Vienna’s University of Medicine, identified six – among a total of 20 – patterns in CT lung images that consistently change in the course of the disease and can be applied to prognoses for the disease. i s p h e r e r i g h t H e m Changes in the brain On the basis of MR images, the researchers studied how the func- tional connectivity architecture in the brain changes in patients with epilepsy or a glioblastoma. Although these diseases are based on focal lesions, the complex networks formed by the neurons change in the effected patients throughout the entire brain. ‘These are plasticity mechanisms that are candidates for markers with which structural chang- es can be recognised earlier,’ explained Langs. Initial find- ings indicate that this image analysis of brain function can be so sensitive that these mech- anisms are recognisable even before a lesion is visible in struc- tural MR images. With all these appli- cations the aim is to create a predic- tion model that makes it possible to predict the future course of the illness. ‘However if a treatment deci- sion is made based on such a model, it must be clear what the underlying mechanism is,’ Langs emphasises. Machine learning is ‘agnostic’, as the computer scientist puts it: in a mass The changed connectome in patients with epilepsy of the left temporal lobe. There is a global change in both the speech network (orange) and, to a lesser extent, the default- modenetwork (blue). The lines show the sub-network where a reduction in connectivity arises in connection with the illness. (Figure: Karl-Heinz Nenning) 20th celebration for web-based PACS originator An ever-advancing portfolio This is a 20th anniversary year for Fujifilm’s Synapse, the world’s first web-based PACS. Today, Synapse 3-D offers advanced 3-D rendering in the Synapse PACS Viewer to perform fast, accurate extractions, stenosis measurements, brain perfusion CT, MRI, and more, the company reports. The Fujifilm Healthcare IT plat- form includes a comprehensive medical informatics and enterprise- imaging portfolio, as follows: Synapse 5 is the company’s next generation PACS. ‘Synapse is one of the fastest medical imaging solutions in the industry, offering sub-second delivery of extremely large data- sets,’ the firm reports. ‘Its underlying architecture promotes significantly less bandwidth consumption and tighter security.’ application Synapse VNA is the most secure, comprehensive for ingesting, storing and providing access to the complete imaging record, the company explains. ‘It securely integrates more specialties, more devices, and more data than any other VNA.’ Synapse Mobility Enterprise Viewer uses the latest server-side render- ing technology to stream imaging securely and quickly to any author- ised user, the company adds. ‘It can be used within applications, directly from your EHR, or on our mobile device apps. Both within and out- side of the Enterprise, giving access to imaging immediately and helping clinicians making the most informed and accurate decisions.’ Fujifilm is at ECR 2019 Expo X5, Booth 503 Synapse 3-D accesses multiple advanced visualisation process- ing tools (more than 50 modules) across multiple specialties including radiology, cardiology, surgery and more. ‘Full integration with Synapse PACS means one-click extremely fast image processing from any Synapse client,’ Fujifilm adds. The clinical workflow manager Synapse CWM advanced radiol- ogy information system continues to evolve, the manufacturer reports. ‘One platform can support acute care facilities, imaging centres, and radiol- ogy practices providing distributed diagnosis.’ Syncro-Dose monitors and man- ages patient radiation exposure across different imaging modalities and facilities. Under its REiLI brand, Fujifilm is developing artificial intel- ligence (AI). Region Recognition, for example, is an AI technology that rec- ognises and extracts organ regions, regardless of deviations in shape, www.healthcare-in-europe.com Associate Professor Georg Langs (Dipl- Ing) studied mathematics and computer science in Vienna and Graz. Following years of research at the Ecole Centrale in Paris and the Massachusetts Institute of Technology (MIT) in the USA, the computer scientist returned to Vienna’s Medical University (MedUni Wien), where today he heads the Computational Imaging Research Lab (CIR, www.cir.meduniwien. ac.at). He also teaches at the University of Vienna, is a reviewer for several international specialty journals, including IEEE Transactions on Pattern Recognition and Machine Intelligence and IEEE Trans- actions on Medical Imaging, and is the author of numerous technical articles. of examples it recognises a pattern that changes consistently and uses this – without regard for whether this pattern coincides with familiar physiological processes. If such a pattern is identified, then the com- puter scientist passes the ball back to biological research. ‘Meanwhile, this Ping-Pong game between radiologists and machine learning experts works very well and leads to progress in the understanding of both sides’, Langs points out. AI is a black box scientists are also trying to manage the so-called ‘Black Box problem’ – that it is impossible to trace from outside how a program based on Continued on page 6 presence or absence of disease, and imaging conditions. Computer Aided Detection is an AI technology to reduce the time of image interpre- tation. Workflow Support, uses AI technology to prioritise study, alert communications of AI findings, and report population automation. * Fujifilm’s artificial intelligence software is a work in progress and is not commercially available in Europe, the company confirms.

LL Communicationwww.fujifilm.eu/ecr2019 FUJIFILM Europe Medical SystemsShaping the future of radiology with advanced imaging technologies.Come and experience FUJIFILM’s advancements in clinical imaging portfolio and IT solutions, and learn how to revolutionize radiology workflow to maximize productivity and ensure a better patient care.Visit us for live demonstrations and subscribe on myESR page to our digital mammography workshops to gain deeper knowledge on the latest imaging applications.FUJIFILM Medical Systems Europe | Booth #503, Expo X5 (Ground Level)Digital Radiography | Women’s Health | Healthcare IT | Artificial Intelligence | Ultrasound

E H @ E C R 9 Distributed learning could be the way forward Radiomics on tap in 5-10 years Keeping data within the hospital by sending the learning modules to each hospital database might prove a game changer in radiomics, a leading Dutch researcher will demonstrate at ECR 2019 Philippe Lambin is a clinician, radiation oncologist and pioneer in translational research, with a focus on hypoxia and Decision Support Systems. He has a PhD in Radiation Biology. In 2016 and 2018 he was an ‘ERC advanced & ERC PoC grant laureate’. He is also co-author of more 450 peer reviewed scientific papers (Hirsch Index) Google scholar: 86. A co-inventor of more than 18 patents, of which five are in the (pre) commercialisation phase, he has also co-promoted over 50 completed PhD’s. Lambin is among the creators of ‘Radiomics’ (animation: http://youtu.be/ Tq980GEVP0Y) and ‘Distributed learning’ a revolutionary Big Data approach for healthcare (http://youtu.be/ZDJFOxpwqEA, visit www.eurocat.info). team at Oncoradiomics, a Maastricht University spinoff, have developed such a solution, which they have called ‘distributed learning from fed- erated databases’. ‘The benefit is that hospitals keep their databanks within their systems, which are protected by firewalls, rather than centralising the multiple databases from different countries,’ he said. The researchers already tested dis- tributed learning at various locations across Europe and the USA, with results as good as software centralis- ing data. Another condition for radiomics to be implemented in clinical practice is to use quality research. A quality score to assess studies is available on the radiomics.world website, which guarantees enough patients are involved without external validation. Lambin recommended using the TRIPOD classification to assess qual- Continued on page 11 attractive. Also, a biopsy may not represent a tumour’s evolution over time. All these issues disappear with radiomics,’ he said. In the meantime, there are still major obstacles to its implementa- tion in clinical practice, and radiom- ics algorithms must be rigorously validated and fulfil an unmet clinical need to pay off. ‘Radiomics, today, is in a chasm – the valley of death,’ Lambin said to describe the gap between scientific validation and applications in clinical routine. Overcoming data confidenti- ality legislation To be adequately trained, algorithms must be fed huge amounts of data from millions of images from differ- ent countries. Collecting these imag- es and sharing the data is an issue, with all the existing data confidenti- ality legislation. Sending the learning modules to each hospital database, instead of having the hospital send their data into a centralised system, would enable that difficulty to be over- come, Lambin believes. He and his Frank J Lexa MD MBA FACR is an academic neuroradiologist and currently a professor and Associate Chief and Vice Chair of Strategy and Leadership, in the Department of Medical Imaging at the University of Arizona. In his book ‘Leadership Lessons for Success in Health Care’ he takes a systematic approach to developing medical leadership skills. ‘If rads don’t face it, non-radiologists will. The choice for us, is change or be changed, disrupt or be disrupted, lead or be led,’ Lexa advised. Radiologists should make the most of disruptive technology. The opti- mistic solution will be to leave the sorting out of cases and eventually the boring cases to the computer, so that radiologists only read the inter- esting ones. ‘A computer that ensures I don’t make a mistake because it has high sensitivity, I would like that, particularly if it’s late at night and if I’m working too many cases that day.’ Rads can ‘merge’ with computers to become ‘centaur radiologists’, a combination that would help them have the best of both worlds. ‘It’s the notion where you have a horse’s body and a human’s upper body, a computer and a radiologist working together to read the images.’ Building value will prove essential to the task, and so will the attitude, Lexa believes. ‘The future belongs to the brave.’ (MR) Radiomics, a field that aims to extract large amounts of quantitative fea- tures from medical images using data-characterisation algorithms, is a major advance for healthcare, according to Philippe Lambin, a radiation oncologist from Maastricht University. ‘The information delivered by radi- ology, and more generally health- care, is still based on a qualita- tive and semi-subjective assessment. There aren’t that many quantitative data in the radiology report,’ he pointed out. Radiomics will help to remedy this situation by enabling the extraction of quantitative and measurable infor- mation on intensity, shape and tex- ture of tumours, plus wavelengths, and complex semantic features that can represent frictions, for instance contact between a tumour and the bone. 15,000 to 20,000 quantitative image characteristics can be gener- ated with radiomics to show gene mutations, differentiate aggressive from non-aggressive tumours, and assess treatment response. The his- tological type of the tumour can also be determined by mining biological information from a CT scan. Within five to ten years, every hospital will use a radiomics solu- tion, Lambin believes. ‘A genomic signature means taking and sending a tissue sample to a lab. Extracting information from an already existing image is cheaper and much more José María Benlloch Baviera is Professor of Physics at the National Spanish Re search Council (CSIC) and Director of the Institute for Instrumentation in Molecular Imaging (I3M) in Valencia, Spain. He worked at the Massachusetts Institute of Technology (MIT), under the aegis of 1990 Physics Nobel Prize winner Jérôme Friedman. Benlloch has published more than 290 articles in scientific reviews, coordinated around 30 investigation pro jects and holds 15 patents. He has also created three spinoff biomedical engineering companies. Professors José Benlloch (left) with Jerome Friedman, 1990 Physics Nobel Prize winner and his former mentor at MIT The EXPLORER total-body PET scanner, released in 2018 and origi- nally conceived by Simon Cherry and Ramsey Badawi from UC Davis, is an inspiring advance. ‘The Explorer has produced exceptional images. I have never seen such high-quality images in a commercial setting,’ Benlloch confirmed. Chinese radiologists have widely purchased the scanner, which enables image-capture of the human body in under a second. However, Western counterparts have not shown the same enthusiasm yet; they tend to prefer big scanners that can accom- modate every patient, even though these are not cost-effective enough for private practice, Benlloch believes. ‘We are going to develop something to interest everyone, i.e. a scanner that can accommodate every patient and has a high turnover – imaging many patients in an hour.’ Having a solution that optimises transcranial ultrasound propaga- tion for blood-brain barrier opening can be an interesting development for brain applications. Benlloch and team recently patented such a device, which works by applying high-inten- sity focused ultrasound (HIFU) to a membrane created with a 3-D printer using CT or MR images. The differ- ence with HIFU is that the solution only uses one ultrasound transducer. ‘We mold the crane based on infor- mation from a CT or MR scan. We cre- ate the membrane, with a 3-D printer, to highlight which brain area we want to focus on with ultrasound. This could be helpful for neurostimulation,’ he pointed out. MRI is excellent to image soft tissues but not so much for hard tissue, such as teeth or bone. The MRI lab at I3M therefore developed MR technology for simultaneous visualisation of both soft and hard tissue in dental applica- tions. ‘Teeth produce a signal on MR, provided one uses a sensor that is fast and sensitive enough to capture that signal. Our solution may be even bet- ter. We change the magic angle of MR by rapidly oscillating between tissues.’ The next frontier will be to image the bone, jaws and crane indifferently, and MR has already proven it could get there using special sequences. The I3M has also given birth to innovations in MR image-guided ther- apy, to drive magnetic nanoparticles to biological tissues using external magnetic fields and guided by MR; and vortex elasticity imaging, using acoustic vortices. The institute has also pioneered MRI development at the cellular range, with non-invasive visualisation of individual human cells in vivo (< 10 μm) and in real-time, and in vivo mouse brain imaging with 29 μm resolution at 15.2-Tesla. Last but not least, the group is also active in software and biomarkers development. www.healthcare-in-europe.com NEW Accu-Gold TOUCH Get in TOUCH with Radcal ! Stand-Alone Diagnostic Radiation Test Instrument • Easy to read 5 inch touchscreen for the X-ray service industry • Access to the full lineup of Radcal Accu-Gold sensors • Simple to use - accurate, reliable, economical • Wired or wireless computer interfaces Visit us at ECR, Booth 138 Call Us 626-357-7921 sales@radcal.com • www.radcal.com

E H @ E C R 1 1 Predicting the truth from hybrid imaging Holomics is a trendy but complex topic When Irène Buvat gained her PhD in Particle and Nuclear Physics from Paris Sud University, France, in 1992, the physicist oriented her path towards nuclear physics for medical imaging applications. After a year at University College London, UK, and two more at the National Institutes of Health, Bethesda, USA, in 1995 she joined the French Centre National de la Recherche Scientifique. Buvat now heads the In Vivo Molecular Imaging research lab at the Service Hospitalier Frédéric Joliot PET centre in Orsay, where her research focuses on developing quantification methods to make the most of SPECT and PET data coupled with CT or MRI, to understand biological mechanisms for optimising patient management. She is involved in radiomic and holomic approaches to further enhance the role of PET/CT and PET/MR in precision medicine. Ever active in training students and knowledge and software dissemination, she promotes open-access to high- standard research material. trends or structures in any given data. Often, clusters can be identified, i.e. groups of data that are similar within each group and different between groups. This approach is very useful because it might reveal trends that we are completely unaware of and might give rise to new hypotheses. ‘In reinforcement learning, we do not provide the algorithm with exam- ples of images associated with the right class they belong to, but we give feedback in the form of reward or penalty. This approach mimics the way humans learn. ‘Therefore, the overall current chal- lenge is to make the most of these dif- ferent approaches to integrate medi- cal images and associated data, and produce new knowledge that will help the medical doctors to optimise patient management.’ ‘Is it possible to know whether a treatment will work before even starting it – in other words, to predict the truth? That’s the great promise of holomics, a concept that everyone has been involved in without even notic- ing,’ said leading French physi- cist Irène Buvat, from the In Vivo Molecular Imaging French lab, who is set to focus on this subject at ECR 2019 (Vienna. 27 February to 3 March 2019). Exclusive interview by Mélisande Rouger ‘The truth,’ said physicist Irène Buvat, discussing whether a therapy predic- tion could work, ‘can involve sev- eral things: the patient’s response to therapy, the prognosis, patient tumour molecular subtype, dementia type, the presence of a genetic mutation – usually something that’s clinically or biologically relevant. We call it truth because we want to make a prediction that will prove to be 100% accurate.’ To predict the truth, one has to account for a number of variables, such as abnormalities seen in blood samples, genetic mutations, the patient’s age, condition, medical his- tory, etc. And of course, imaging data, which are very useful and currently absolutely needed for patient manage- ment. This is especially true for hybrid imaging data, Buvat pointed out. Prediction of response to neo-adjuvant therapy from baseline scans in breast cancer patients: the holomic model yields the best sensitivity and specificity ‘In contrast to current predictive models, holomics includes all these different types of information to predict the likelihood of a patient responding to therapy, developing a certain disease, and so on. The assumption is that information reflect- ed by images will tell us more if they are put into context, i.e. if interpreted together with other non-imaging data.’ What are holomics? ‘Holomics can be defined as the gath- ering of genomic, radiomic, proteom- ic, clinical, immunohistochemical and many more data, and their integra- tion in predictive or prognostic mod- els. Holomics can be seen as an exten- sion of radiomics, which is the extrac- tion of features from many images as a foundation for predictive models. In short, holomics is radiomics enriched by other types of data. What are the potential benefits? ‘Being able to predict will definite- ly contribute to precision medicine, because if we can predict that a patient has a very low likelihood to respond to a given chemotherapy and there is an alternative treatment, that treatment will be given first. ‘Patient management also heavily depends on the experience of the specialists involved. The idea with holomics is to erase this difference by Radiomics on tap in 5-10 years Continued from page 9 ity of the biomarkers used in a trial. ‘Radiomics signatures must meet the TRIPOD level 4 to be deemed wor- thy. Prospective studies may also be required, with a predetermined signature that needs to be validated in the study,’ he said. Software must also receive CE mark or FDA approval. Integration into the hospital workflow is key for the solution to bring value. Last but not least, reimbursement can be problematic when financial incen- tives go in the wrong direction. ‘Using our radiomics signature, we can reduce the number of useless interventions in kidney cysts, for example, but in some countries hos- pital managers argue that they are paid per intervention,’ he said. Combining the best of DL with radiomics An interesting path to explore in the future is the alliance of radiomics and deep learning (DL), notably to develop non-invasive imaging based biomarkers for radiomics. ‘Current www.healthcare-in-europe.com research combines the best of DL, in particular automated segmentation, with the best of radiomics, by enforc- ing robust image characteristics into DL,’ Lambin pointed out. ‘This looks like a promising approach.’ DL uses a radically different method than radiomics, in which image characteristics with an already known image signature, for instance heterogeneity, are selected. So far no one can explain how DL works exactly. ‘It’s a black box; we don’t really know how it works , which is annoying because doctors like to understand the process,’ he said. DL also requires more horsepower to train its algorithms to distinguish between a malignant and a benign tumour. To feed the beast, DL needs at least 10,000 images and, some- times, synthetic data. ‘If we could have a million data, I bet that DL would be better than traditional radi- omics. But,’ he concluded, ‘we would still have the issue of an uninterpret- able algorithm.’ (MR) ‘We use both shallow and deep learn- ing, which differ in the way infor- mation is extracted from the data. To design a model, three types of learning techniques can be employed: supervised, unsupervised or rein- forcement learning; supervised and unsupervised being by far the most frequent now. ‘As an example for supervised learning, we tell the algorithm which tumours in a data set are benign and which are malignant. Then, the machine learning approach will find the model that makes the most accu- rate prediction for the data present in the training set. ‘Unsupervised learning, on the other hand, is performed to try and identify integrating different kinds of informa- tion on a quantitative and reproduc- ible basis to assist therapeutic deci- sions. Any decision would of course be made under the control of a spe- cialist, but the algorithm would always provide expert knowledge. ‘ Where is it interesting to use hol- omics? ‘Holomic approaches are mainly being developed in oncological imaging, but are also valuable in cardiovascular and brain imaging. Right now, people cre- ating scores involving image-derived parameters and other patient features like age are doing holomics without noticing. The word was first coined in the context of tumour imaging. ‘Holomics is a trendy, but complex topic, he pointed out. ‘We have a few results combining blood markers with imaging markers, but we could have far better results if we could add information from genomics as well as other fields.’ Why is it appealing to implement holomics with hybrid imaging? ‘The more different the imaging data, the deeper we can use holomics. Hybrid imaging gives information from two imaging modalities in the same machine at the same time, which makes it a privileged gateway to col- lect many distinct types of informa- tion. PET and MR offer complemen- tary but non-redundant information that can give a more complete profile of the disease. We can see anatomic characteristics on MR and, at the same time, metabolic features on PET, and additional functional features using different MR sequences. ‘The main obstacle now is to col- lect enough reliable and rich data to build holomic models. There are fewer hybrid imaging modalities, but the information they provide is richer than CT or MR scanners alone. In hybrid scanners, images are perfectly aligned in space and time, unlike two exams performed on two different machines at different times.’ How exactly do holomics work? ®www.sternmed.deWe make it possible

E H @ E C R 1 3 Future magnetic resonance scanners MRI speed and versatility raised by AI actionable information. Similarly, AI can take the information from con- tinuously acquired datasets and can then reconstruct full images from partial data. Snapshots are dead, long live streaming As part of his work, Sodickson and the NYU School of Medicine’s department of radiology is collabo- rating with the Facebook Artificial Intelligence Research (FAIR) group to speed up MRI scans while still acquiring enough data. He sees the work developing in two phases — using existing tech- nology to gather less data and then innovating scanners to operate dif- ferently, almost like a self-driving car with multiple sensors. ‘The days of the carefully framed snapshot are, if not already over, certainly limited,’ Sodickson said. ‘It behoves everyone to think about how to construct data streams rather than image series. In MRI, and imag- ing more broadly, we’re not just following our eyes anymore; we’re emulating how we see the world. Stay tuned for imaging devices that start looking and feeling radically different than we’re used to.’ The Facebook AI research cooperation At the end of November 2018, the NYU School of Medicine’s Department of Radiology released the first large-scale MRI dataset of its kind as part of the fastMRI project launched earlier that year with Facebook Artificial Intelligence Research. While other sets of radio- logical images have been released Dr Daniel Sodickson is professor and vice chair for research in the radiology department at NYU, and a principal investigator at the Center for Advanced Imaging Innovation and Research. He also a director at the Bernard & Irene Schwartz Center for Biomedical Imaging. Having gained a BSc in Physics and BA in Humanities from Yale College, he received his PhD in Medical Physics from MIT and MD from Harvard Medical School, both as a part of the Harvard-MIT Division of Health Sciences and Technology. Sodickson’s research primarily focuses on developing new techniques for biomedical imaging, with the broad aim of seeing what has previously been invisible. He is credited with founding the field of parallel imaging, in which distributed arrays of detectors are used to gather magnetic resonance images at hitherto inaccessible speeds. previously, this dataset represents the largest public release of raw MRI data to date. The first phase of the project will involve data from knee MRI scans, but future releases will include data from liver and brain scans. ‘This collaboration focuses on applying the strengths of machine learning to reconstruct high-value images in new ways. ‘Rather than using existing images to train AI algorithms, we will radi- cally change the way medical imag- es are acquired in the first place,’ Sodickson explained. ‘Our aim is not merely enhanced data mining with artificial intel- ligence, but rather to create new capabilities for medical visualisa- tion, to benefit human health.’ the ‘MRI in five minutes – Dream or Reality?’ That’s question posed and answered by Dr Daniel Sodickson during the International MRI Symposium held recently in Garmisch-Partenkirchen, Germany (17-19 January). Sodickson, from the New York University School of Medicine, sees a number of changes entering imag- ing, including artificial intelligence (AI), which will make MR imaging much faster and more versatile. ‘I see the five-minute MRI as a sort of bellwether for trends in imaging today,’ Sodickson said. ‘AI might enable a new parable of ultra-fast MRI that’s going to change the expe- rience of the patient and day-to-day workflow of the radiologist. There will be a number of other changes that go along with AI to make imag- ing super-fast.’ An alternative title for his talk might have been: Imaging in a changing world: the scanners of the future and the future of scanning. ‘In the future,’ he predicted, ‘MR scanners will move from snapshots to streaming.’ A flexible MRI detector array in the form of a glove. Top: photos of the hand in various poses. Bottom: corresponding MR images obtained using the glove. With multifaceted sensor technology like this, MRI can embrace new flexibility. (Credits: Bei Zhang, Martijn Cloos, Daniel Sodickson. Zhang B et al, Nature Biomedical Engineering 2018;2(8):570-577. A high-impedance detector-array glove for magnetic resonance imaging of the hand.) the brain’, and asks, ‘How do we make sense of all that information? The model can come from how our own brains work.’ All our senses — hearing, sight, smell, touch — are filtered through the brain, which turns them into AI as the brain of MRI Traditionally, clinicians devote their efforts to one view of the anatomy, or a series of slices, or dynamic view, pausing between shots to carry out the examination setup. In the future, Sodickson predicts that software will adapt to a more modern paradigm, in which the scanner gathers information about the patient continuously and the software then pieces the images together. Sodickson likens this to a ‘move from emulating the eye to imitating c i g o o H l : e c r u o S Pete Valenti is Hologic’s division president of breast and skeletal health solutions the concept from AI mammogra- phy to the world of ultrasound – working in partnership with Clarius - whilst continuing to expand its expertise within the breast diagnosis and therapeutics field. In terms of tomosynthesis and 3D mammography, he said Hologic is driving innovation faster than any other organisation and that is pri- marily because of the clear focus on breast care. As for the future, he said an area of development is increased port- ability and also looking to integrate processes as well as the ability for health systems to better access patients away from a hospital set- ting. The aim, concluded Valenti, is to deliver improved outcomes and workflow as well as increase patient satisfaction as Hologic drives inno- vation in breast health technology. www.healthcare-in-europe.com NEWTOM 5G XLEXTRA.VISION www.newtom.itCUTTING EDGE PERFORMANCE FOR ADVANCED DIAGNOSTICS.Advanced diagnostics with 5G XL, the only CBCT device with lying down patient positioning that offers excellent stabilisation and a broad range of FOVs for very high quality 3D and 2D images. 5G XL is the first device with native FOV 21 x 19 cm for highly detailed investigations. The ideal device that produces clear, high definition images for orthopaedics, otorhinolaryngology, maxillofacial surgery and dentistry applications.SUPERIOR DIAGNOSTIC QUALITY Very high resolution 2D and 3D images with a broad range of FOV for an extensive selection of clinical applications.OPTIMAL LYING DOWN POSITION The only CBCT system with patient lying down positioning, a motor-driven patient table and an open gantry. Perfect patient stabilisation considerably reduces any movement-induced artifacts.LOW X-RAY DOSE The ECO Scan mode and SafeBeamTM technology offered by 5G XL further reduce the dose radiated to the patient, compared to examinations performed with CT technology.SPECIALIST SOFTWARE The adaptive user interface displays images and provides access to innovative 3D and 2D analysis functions for rapid and accurate diagnoses and optimal workflow.

E H @ E C R 1 7 Dementia diagnosis MRI is the first step the case in other countries, Hess said. Some countries use CT, or no imaging at all, to examine such patients, so he found it interesting to hear discussions from delegates on how MRI, and imaging in gen- eral, is used for dementia cases in other parts of the world. Hess believes that radiologists can play a significant role in diagnosing dementia and that MRI exams of people with cognitive complaints should be reviewed differently from patients showing with other issues. ‘In particular,’ he emphasised, ‘it requires a sound understanding of the neuropathology and neuroanat- omy of dementia.’ Dr Christopher Hess is the Alexander R. Margulis Professor and Chairman of the Department of Radiology and Biomedical Imaging at the University of California, San Francisco. He gained his PhD in Electrical Engineering from the University of Illinois at Urbana Champaign. Clinical residency and fellowship followed at UCSF. He is co-Deputy Editor of Neuro- radiology for Radiology; on the American Journal of Neuroradiology, editorial board and a member of the Radiological Society of North America, the International Society for MR in Medicine, the American Society of Neuroradiology, and Academy of Radiology Research, and he is a fellow of the American Inst. for Medical and Biological Engineering, Research focus: brain degeneration, epilepsy, and neurovascular disease. i t n a D a e r d n A / k c o t s r e t t u h S © areas of the brain that alter cogni- tion, Hess pointed out. He then discussed how radiolo- gists can use these findings to sup- port or refute the diagnosis of spe- cific neurodegenerative processes, as well to recommend appropriate next steps in disease evaluation and management. He shared the approach he takes with other clini- cians. ‘Brain MRI,’ he said, ‘is often the first step in evaluation. General radi- ologists don’t necessarily recognise the important findings related to dementia. ‘We will review dementia symp- toms and how they should guide the eyes of radiologists, and we’ll look carefully at the critical importance of distinguishing between rapidly progressive and chronic dementia.’ Hearing how MRI, or other imaging, is used in demen- tia cases around the world While MRI is the standard of care in the United States in the initial pres- entation of patients with cognitive complaints, that’s not necessarily ) o c s i c n a r F n a S , a i n r o f i l a C f o y t i s r e v i n U © www.healthcare-in-europe.com

E H @ E C R 1 9 Multidisciplinary care is key to cardiac disease management Research with 7-Tesla MRI New 7-T MR methods could potentially shed light on cardio- myopathies’ principles, according to a leading French radiologist who also stresses the importance of teamwork between radiolo- gists, cardiologists, surgeons and anaesthesiologists. Morphologic and dynamic informa- tion of the myocardium is achieved with millimetric resolution (0.9x0.9 mm2). Strong intensity variations characteristic of 7-Tesla MRI can be observed from anterior to posterior myocardial segments. New tools provided by indus- trial partners and used by cardio- vascular surgeons and radiologists are improving treatment of thoracic aorta pathologies. An increasingly used technique is fusion imaging, in which pre- treatment MR and CT scans of the patient are being fused with angi- ography images to guide stent-graft navigation through the vascular structures of the patient during the intervention, according to Alexis Jacquier, cardiovascular radiologist at Timone University Hospital in Marseille. ‘Fusion imaging enables to lower radiation dose and to reduce the amount of contrast media that are traditionally required in this type of surgery. It avoids having to inject iodine to know where we’re at,’ he explained. Covering full aortic patholo- gy management, from diag- nosis to care and follow-up The hospital also hosts the Timone Aortic Centre (CAT in French), a leading regional multidisciplinary centre that covers full aortic pathol- ogy management, from diagnosis to patient care and follow-up, with a strong connection with the univer- sity. The CAT includes vascular and cardiac surgeons, radiologists, car- diologists, vascular physicians and anaesthesiologists. The objective is to provide a mul- tidisciplinary approach to provide the best medical care; for instance all thoracic stent-graft procedures are performed by a multidiscipli- nary team comprised of vascular surgeons, radiologists and anaesthe- siologists at CAT. Figure 1 Figure 2 Sodium mapping of the heart using 7-Tesla MRI Higher signal-to-noise ratio (SNR) using 7-Tesla MRI allows to map sodium in the human heart. The isolation of the long relaxation sodium component (using a long echo time) and the compen- sation of concomitant signal modulation (T2star and B1+/B1- from transmit-receive coil) allows for a single slice to be mapped within 5min at a resolution of 2x2x10 mm3. Reference sodium concentration vials attached to the coil serve as calibration of the sodium MRI signal. Figure 1: Top-left is B1+/B1- map, Top-right is T2star map, Bottom-left is anatomical MRI image of the heart using conventional 1H-MRI, and bottom-left is the map of long- relaxation sodium concentration after corrections of the sodium MRI signal using the maps at the top. Figure 2: Overlay of long-relaxation sodium map onto the anatomical MRI. Imaging has become key in tho- racic aorta treatment with the boom of minimally invasive procedures. Besides thoracic disease, Timone Hospital is one of the main cen- tres in France offering endovascular interventional radiology skills to treat patients with carotid and renal disease, which Jacquier and col- league Vincent Vidal perform daily, along with the full suite of car- diovascular interventional radiology procedures – endoprostheses and stent placement, small vessels and tumour embolisation, etc. Furthermore, the hospital is locat- ed close to the medical and biology MR centre (CRMBM), one of the few laboratories in Europe that work with 7-T MRI for diagnostic imaging research. This proximity enables Jacquier and team to test 7-T methods using sodium instead of proton imag- ing, a possibility that opens brand new perspectives in heart imaging. ‘Sodium electrolytic disorganisation in the myocardium can have an electrical and mechanical impact on heart function. 7-T will enable the development of new applications in the field. It is still a complex task, but we are working hard on different papers on sodium quantification in the myocardium and potential clinical applications’ he explained. Cooperation with cardiologists is essential in myocardial disease man- agement, according to Jacquier, who again stressed the importance of the multidisciplinary approach during patient treatment. ‘Patients are now being cared for within the heart team, a model increasingly followed by healthcare facilities in France and beyond,’ he explained. ‘Whether it’s for TAVI procedure, diagnosis or follow-up. Medicine is becoming hyper specialised and mixing profiles and specialties ena- bles us to significantly improve patient care.’ Obligatory three-step train- ing for residents in emer- gency radiology Another significant development in France was the reform of the radiolo- gy residents’ training scheme, which was introduced in 2017. Radiology residents must now undergo a three- step training, including successively: base training (one year), dedicated Cardiovascular radiologist Professor Alexis Jacquier, at Timone University Hospital, Marseille, France, trained in Marseille and Lyon and gained his PhD in San Francisco, USA, supervised by Maythem Saeed and Charles Higgins. In 2006 he integrated the cardiovascular group in the CEMEREM research lab (http://crmbm.univ-amu.fr). He is author and co-author of more than 90 peer-reviewed publications and has presented numerous lectures, tutorials and refresher courses internationally. He also chaired the European Society of Cardiac Radiology membership committee and is current vice president of the French Society of Cardio-Vascular Radiology (Société Française d’Imagerie CardioVasculaire, SFICV). prescribes the CT and MR scans and radiologist performs the tech- nical assessment and writes the report – and then sends it to the cardiologist. Jacquier: ‘This division of tasks promotes the best possible medical care, but everything really depends on the physician’s skills. A lot of things may need to be updated as we gradually introduce artificial intelligence.’ Radiologists must also homoge- nise the way they write the imaging report. Introducing the structured report to exploit data at national level will prove essential for their future. Another priority is to improve communication not only with patients but also other medical spe- cialties, he said. The radiologist is no longer alone in a basement reading scans,with no contact Jacquier will participate in the International Day of Radiology (8 November 2018), an initiative to highlight the radiologist’s role in cardiac care. ‘Radiology is not a medico-tech- nical specialty, although French administration still classifies us as such. We’re a medical discipline. The old-fashioned image of the radiologist reading scans alone in a basement and not having contact with anyone else in a hospital is outdated. ‘The radiologist,’ he emphasised, ‘is now at the centre of patient care and healthcare.’ (MR) to emergency radiology; in-depth training (three years), to ensure that every subspecialty in radiology has been covered in their education; and consolidation training (one or two years), providing certification for one or two subspecialties. The French Council of the Teachers of Radiology (CERF) has been piloting the change for radiology. The French Society of Cardiovascular Radiology now pro- vides e-learning material to ensure homogeneous teaching and training program across the country. Last September, the series became freely available for French residents on the CERF website, and also available for all radiologists on the website of the French Society of Radiology. This change is a substantial improvement in the training scheme, because it reflects daily routine bet- ter, Jacquier added. ‘Cardiac imaging studies are being prescribed every day by all sorts of physicians: GPs, endocrinologists, surgeons, and even oncologists, for instance in pre- and post-chemotherapy evalu- ation.’ Many things may need updatewing for AI As for cardiology, the French Society of Cardiology and the French Society of Radiology estab- lished a working protocol in 2005; according to this, the cardiologist High-resolution Simultaneous Multi-Slice (SMS) dynamic MRI of the heart at 7-Tesla Increased signal-to-noise ratio from the 7-Tesla MRI is harnessed for refined imaging of heart. Simultaneous Multi-Slice (SMS) cardiac dynamic MRI (cine) permits the acquisition of three thin- slices (4 mm) within a 10 s breath-hold. Robustness to patients motion and limited breath-hold capacity is guaranteed through a dedicated self-calibrated SMS technique tailored for cardiac imaging. Morphologic and dynamic information of the myocardium Figure 3: SMS cine acquired within 10s showing diastole (relaxed phase of the cardiac cycle) of the apex (a) mid- ventricular (b) and base (c) slices. is achieved with millimetric resolution (0.9x0.9 mm2). Strong intensity variations characteristic of 7-Tesla MRI can be observed from anterior to posterior myocardial segments. a b c www.healthcare-in-europe.com Visit us at ECRHall X5Booth 532Next generation medical imagingwww.jvc.eu

E H @ E C R 2 1 Artificial intelligence themes dominated RSNA2018 Radiologists reveal optimism about AI orbital wall fracture surgery can sig- nificantly reduce operating time. Findings from a study conducted at the hospital showed that, when such a model was used, surgical pro- cedures were reduced from an aver- age 96 minutes to 64 minutes. Nicole Wake PhD, of New York University School of Medicine, discussed how 3-D printed prostate models created from pre-operative MRI data can be used to correlate MRI and pathology and also help patients better under- stand their disease and treatment. A nearly 60% vendor increase in the Machine Learning Showcase compared to 2017 was notable. This included numerous theatre lecture presentations, 22 interactive kiosks, and 1,496 square meters of exhibit space for 78 participating compa- nies. An RSNA representative told European Hospital that the spe- cialised showcases represent the organisation’s interest in emerging technologies. ‘Our members have expressed interest in learning more about machine learning, AI and 3-D printing and advanced visualisation products, including augmented and virtual reality platforms. These tar- geted showcases represent a one- stop destination for our attendees to see and learn more about these products in an integrated setting.’ As always, clinical topics were numerous, diverse and impressive. A huge educational exhibit centre in a dedicated hall, and technical prod- ucts exhibited by 732 companies in 40,300 square metres, provided the usual overwhelming data over- load. Although attendance has not matched that of 2014, online digital access to ‘virtual sessions’ expanded through March 2019, reaches radi- ologists worldwide. co-director of the University Hospital of Basel’s 3-D Print Lab and a senior physician in cardiac and thoracic diagnostics, discussed how the use of CT data-based 3-D printed models to pre-contour implants in inferior You can discover these innovations during ECR tradeshow, booth 410 – Foyer D! User experience and the optimisation of daily workflow: at the heart of our developments The professionals we all rely on to keep us healthy, they rely on X-Ray imaging systems empowered by flat panel detectors and software solutions provided by Thales. Our solutions offer superior accuracy and advanced image pro- cessing to help image interpretation while decreasing radiation exposure. They also include a variety of innovative features that simplify the daily workflow of healthcare professionals while capitalizing on new technologies such as Artificial Intelligence, Big Data and the Internet of Things, without compromising on Cybersecurity. Pixium® Portable 3543 DR with Embedded Patient Identification: Portability for Excellence and Efficiency in Hospital Care This world’s first ever patented solution allows for images to be taken at the patient’s bedside thanks to an embedded barcode reader that scans patient ID bracelets. This feature links images to that specific patient and stores them directly on the detector. An easy-reading display indicates the number of images taken per patient and the total number of images stored in the device for an ulterior upload. As one of the most cost effective systems on the market, allowing multiple examinations and eliminating the need for numerous CR cassettes, the Pixium® Portable 3543 DR facilitates the daily work of healthcare professionals. It ensures greater productivity, making it also one of the most attractive solutions available on the market. Dynamic Flat Panels Fully Redesigned for Enhanced Competitiveness Thales is fully engaged in optimising the cost of its flat panel detectors dedicated to interventional applications. Pixium® 2121 & 3030 Efficiency detectors have a new design that matches budgetary requirements. The universal software platform, PixDyn, is ensuring cost effective integration and sup- ports the transition from conventional systems to flat panels for all mobile C-arm systems. The need to deliver unparalleled visual support in real time remains our main focus, with no compromise on image quality. From Detectors to Turnkey Imaging Solutions As well as having world leading experience in flat panel detec- tors, Thales meets healthcare professional’s needs for innovative software solutions that optimise both image quality and daily workflow. ArtPix™ DRF, a new imaging solution for fluoroscopy, radiog- raphy and angiography applications, raises the bar in terms of image quality and versatility. The end-user can easily person- alise features such as the user interface, display configuration, image quality and room peripherals. The customisable system, fully adaptable to hospital workflows, reduces costs by enabling a higher throughput of patient examinations. It offers a vast choice of advanced clinical options such as: Tomosynthesis, DSA (Digital Subtraction Angiography) stitching, radiation less posi- tioning, etc. Our capabilities are taking digital radiology technology to new levels Thales Group considers embedding Artificial Intelligence into its solutions as a key priority. In these developments, we are fully supported by Thales teams at the Digital Factory in Paris, France and R&D specialists at the CortAIx center, dedicated to AI in Montreal, Canada. Our goal is to implement Artificial Intelligence as an aid for patient diagnosis as well as improving hospitals workflow. Additionally, Thales’s cybersecurity expertise is providing hospitals with new data protection methods, shielding personal medical information and avoiding potential data breaches. The needs of manufacturers and healthcare professionals are at the heart of our activities. Thanks to this, we are proud to be part of half of radiological patient examinations worldwide. increase the value of radiology reports, and to enable virtual rounds and virtual consultations when need- ed, Recht also emphasised the need for AI use to improve data and image acquisition in real-time. He cited the benefits of deep learning algorithms with pattern recognition and image reconstruction capabilities to cre- ate quality diagnostic MRI images within five minutes, which could have a dramatic impact on report turnaround time, patient throughput and department workflow. Better data analysis could improve overall department operations, iden- tifying areas for improvement in interpretation accuracy, efficiency and communication. Data correlat- ing diagnoses with clinical pathol- ogy and outcomes could be used in a continuous feedback mode to perform smarter patient imag- ing and improving evidence-based guidelines. A machine learning competition The RSNA also sponsored a machine learning competition to develop algo- rithms that could identify and local- ise pneumonia in chest X-rays. Over 1,400 teams took part in the RSNA Pneumonia Detection Challenge, and 346 submitted results for evalu- ation. Alexandre Cadrin-Chênevert MD, a Canadian radiologist and com- puter engineer at CISSSL of Saint Charles Borromée, Quebec, and Ian Pan, a third-year medical student at Brown University in Providence, RI, created the winning algorithm and received a $12,000 prize. 3-D medical printing was also emphasised in scientific presenta- tions as well as a dedicated 3-D Printing and Advanced Visualisation Showcase. Sixteen companies par- ticipated in a 186 square metres venue, which combined education- al lectures and demonstrations of products and techniques to create 3-D-printed models to improve pre- surgical planning and intra-operative guidance. Efforts are underway in the USA for radiologists to be com- pensated for medical 3-D printing, which is expected to boost utilisa- tion significantly. As just two examples… in a scien- tific session, Philipp Brantner MD, www.healthcare-in-europe.com www.thalesgroup.com

E H @ E C R 2 3 Practical videos, e-booklets, case studies and more lar ultrasound in several vascular conditions. All courses and con- tent of www.abcvascular.com have been accredited by the EACCME (European Accreditation Council for Continuing Medical Education) to provide CME credits. split screens are used to show patient positioning, operator’s hand/ ultrasound transducer movements and ultrasound videos. This meth- od of teaching is essential, as the information to acquire is inherently visual and dynamic. The aim is to offer learners a virtual experience of a one to one learning session with a vascular ultrasound expert. Learners are then challenged with real life case scenarios, where the explained diagnostic criteria are applied and explained. This will give the learn- er a virtual experience of vascu- Online education in vascular ultrasound Fabrizio d’Abate, St. George’s University Hospitals NHS Foundation Trust, in London, UK describes new aspects of learning and training in ultrasound operation the more advanced concepts used to diagnose vascular diseases, sim- plified into key learning points to make learning fast and easy. The courses are supported by commented video-lectures where ditional A textbook represents the most tra- tool of a teach- ing arsenal. However, the I T boom and internet have transformed the way people approach different tasks in their lives, from solving a problem to acquiring knowledge. This has influ- enced the way of learning, that has become more dynamic and interac- tive. Never before has education been available on a large scale as it is now, with so much information and knowledge to convey to learners. Online courses are increasingly nec- essary for education ‘becoming not an add-on feature in teaching but a necessity’. With online education users can proceed through a training program at their own pace, can access the training at any time, receiving only as much information as they need in an interactive way. The number of people opting for online courses worldwide is increasing due to busy lifestyle. Ultrasound is a fast growing diag- nostic technique and is in constant demand, however, there remains a shortage of experienced operators in such a field, especially in vascular ultrasound. The lack of a dedicated e-learning teaching platform in vascular ultra- sound and the common passion in education shared by a vascular sonographer and a cardiovascular medicine doctor, gave life to www. abcvascular.com a new e-learning platform to support healthcare professionals who want to acquire knowledge and skills in vascular ultrasound. The ABCvascular mission is to walk learners through all the steps necessary to learn vascular ultra- sound, from the foundation notions (the ABC of vascular ultrasound), to Ultrasound training exoert Fabrizio d’Abate www.healthcare-in-europe.com

L A B O R AT O RY 2 5 A powerful tool to differentiate opioids Mass Spec detects illicit drug presence As a synthetic opioid approved for treat- ing severe pain, fentanyl has shown clear medical benefits. However, in recent years, continuous abuse of fen- tanyl and its derived analogues sub- stances has become a major public health issue – overdoses and deaths associated with illicitly-manufactured fentanyl rose dramatically. Some of these synthetic analogues have extremely high potency and only require a small amount for accidental overdose (image shows the difference in amount of a lethal dose of heroin vs fentanyl vs carfentanyl). Drug deal- ers have exploited this, using traces of fentanyl to increase potency in other drugs to reduce costs. These analogues are structurally similar to fentanyl, so they often go undetected. Mass Spectrometry (MS) has emerged as a powerful tool to differentiate drugs, and has particular applications in Emergency Departments where speed is often vital. Dr Pierre Negri from SCIEX – which offers a variety of LC tandem mass spectrometry solutions in foren- sic and toxicology analysis – said that because people began buying drugs containing fentanyl from dealers and overdosed, the need grew to devel- op methods to quickly detect small amounts of fentanyl so that first aid responders could take appropriate actions. Where MS offers a solution is in its characterisation of various mol- ecules (including drugs) as a highly sensitive and accurate analytical tech- nique capable of providing qualitative ) ) / v o g . c d c . w w w / / : s p t t h ( n o i t n e v e r P d n a l o r t n o C e s a e s i D r o f s r e t n e C : e c r u o S ( The continuous abuse of fentanyl and its derived analogue substances continues to be a serious public health and safety problem. and quantitative analytical data on extremely small amounts of analytes. ‘As a result, mass spectrometry can differentiate drugs by assessing their mass, or molecular weight, hence providing the necessary information to elucidate their molecular structure,’ continued Negri, who is the SCIEX lead for global technical forensic mar- keting. ‘This technology can be used to detect and quantify fentanyl and its analogues in small quantity while elucidating the chemical structure of these compounds.’ Today, LC tandem mass spectro- metric analysis (LC-MS/MS) is the preferred method of analysis for forensic over traditional techniques, for both screening and conformation purposes. Immunoassays are com- monly used as a first-line screening method abused drugs detection and rely on the presence of a specific antibody or specific antigen in urine, blood or other fluids, yet they lack sensitivity in detecting designer drugs and require additional confirmation, which slows the analytical process. ‘The advantages of LC-MS/MS over conventional methods are that it has better compound detection sensitiv- ity for significantly lower levels than the current cut-off levels; faster time to results, with less sample prep and no derivatisation; saves time and money with multiplexing capabilities to analyse for many compounds in a single analysis; and has the ability to Towards a clearer view of complex lab results simultaneous quantify and identify many drug targets in various sample matrices such as saliva, urine, blood or hair samples. ‘It’s important to note that, as with immunoassay, assay design is crucial for accuracy and specificity with mass spectrometry. However, unlike immunoassay-based detection, mass spectrometry provides analyte specific results and allows accurate quantification of the drugs analysed with far greater sensitivity and resolv- ing power than immunoassay.’ In recent years, mass spectro meters have become more accessible, cheap- er to purchase and operate, and easier to use. Additionally, new MS-based work- flows can be developed to eliminate a two-stage process currently used for drug monitoring, by allowing technicians to run many tests at once with a high throughput, cut- ting processing time while improv- ing screening accuracy with qual- ity assurance. Negri suggests that, at present there remains a lack of reliable and inexpensive tests that allow for comprehensive surveillance of synthetic drugs flooding the illegal market, whereas the development of novel MS-based screening techniques have led to the rapid and accurate identification of fentanyl, as well as other drugs of abuse, previously dif- ficult to detect by traditional analysis. He said the SCIEX X500R QTOF is a powerful tool for forensic researchers investigating samples for unknown Dr Pierre Negri leads global technical foren sic marketing for SCIEX, collaborating with opinion leaders in forensic and criminal toxicology research to develop and implement new mass spectrometry methods. In the labo ratory he also works with customers to develop and promote novel mass spectro metry analysis workflows applied to con trolled substances and trace evidence materials. compounds, drug metabolites, unknown chemicals or hazards, or novel psycho-active substances that have never been previously detected or characterised. Powered by the user-friendly SCIEX OS software, it can redefine how to analyse a rou- tine forensic drug screening sample. The 6500 QTRAP or 6600 TripleTOF technologies can assist with research and method development. ‘The versatility of modern mass spectrometers enables specific and comprehensive measurements that allow drug detection with a high level of sensitivity and specificity,’ he said. ‘The implementation of this tech- nology in drug screening over the years has shown improved accuracy, precision, sensitivity, selectivity, and clinical utility over simple direct and complex reference immunoassays.’ Where seconds matter, Negri added, being able to correctly iden- tify the fentanyl analogue screened, is critical (MN) Shaping post-analytical interpretive tools from some 30 million newborns tested in 69 countries, disease ranges were compiled from >21,000 true positive cases. Interpretation of complex profiles of laboratory results can be sig- nificantly improved via multivari- ate pattern recognition software, Mark Nicholls reports. A multivariate pattern recognition software, developed by a team led by Dr Piero Rinaldo at Mayo Clinic in Rochester, USA, aims to integrate results to diagnose a particular con- dition in a single score, in an objec- tive evidence-based way that is open to worldwide collaboration and data sharing. Currently co-director of the Biochemical Genetics Laboratory and vice-chair of Information Management in the Department of Laboratory Medicine and Pathology at the Mayo Clinic, Rinaldo out- lined the work at the Frontiers in Laboratory Medicine (FiLM) confer- ence held in Birmingham, UK, in January. Explaining how the software pro- duces post-analytical tools available on demand through a web interface, the approach was first applied to newborn screening, and has grown to involve 269 programs in 69 coun- tries. In his presentation ‘Personalised reference ranges; developing indi- www.healthcare-in-europe.com vidualised reference ranges’, Rinaldo suggests there are ‘limitations and pitfalls’ of the traditional way to establish reference ranges and cut- off values for laboratory tests. ‘In my view,’ he said, ‘it’s being done in a manner that seriously underestimates the role of covariates like age at collection and sex. For so many markers, we use the same deci- sion limit for a 21-year-old healthy male and for an 80-year-old obese female as well.’ Moving forward, he advocates the CLIR (Collaborative Laboratory Integrated Reports) approach and the importance of individual refer- ence ranges, paving the way for how lab tests should be routinely ordered in the era of personalised medicine. CLIR, he said, replaces conven- tional reference ranges with continu- ous covariate-adjusted percentiles; replaces analyte cut-off values with condition-specific disease ranges; enhances the clinical utility of indi- vidual markers with all possible per- mutations of ratios, and replaces diagnostic sequential algorithms with tool-based parallel algorithms. However, he stressed: ‘The impor- tant point to recognise here is that creating better reference ranges is just one initial step of the process, not the end point. That’s the post- analytical interpretive tools.’ The basic tenant of CLIR is world- wide collaboration and data sharing to establish covariate-adjusted refer- ence ranges based on millions of data points and disease ranges. ‘The software evaluates the degree of overlap of any marker (after covar- iate adjustment) between the two ranges separately for every possible target condition, and use it to com- pile a cumulative and integrated score. The comparison of this score versus those calculated for known cases offers an objective assessment of the disease likelihood, a vast improvement over the binary choice of normal/abnormal. ‘The other relevant concept is that clinical validation is an evolving pro- cess, as new cases are added (daily) to the database the tools improve accordingly, such as via machine learning.’ CLIR was originally developed to support Region 4 Stork (R4S), a USA federally-funded collaborative project that started in 2004 aimed at improving newborn screening by tandem mass spectrometry. Since 2012, R4S and CLIR online tools have been utilised more than half a billion times, with an average of more than 200,000 scores calcu- lated daily worldwide. In recent work showing the first prospective application of the sec- ond generation of the software, the team focused on the implementation of newborn screening for lysosomal disorders, an area where there has been overall poor specificity, psycho- social harm experienced by caregiv- ers, and costly follow-up testing of false-positive cases. interpretive The study concluded that: ‘Post- analytical tools can drastically reduce false-positive out- comes, with preliminary evidence of no greater risk of false-negative events, still to be verified by long- term surveillance.’ While the outcome of the R4S pro- ject was a false positive rate (FPR) of 0.024% and a positive predictive value (PPV) of 69% (US average ~20%) while the first application of CLIR to three lysosomal disorders achieved a FPR of 0.001% and a PPV of 87% and has led to the discovery of a novel biomarker. In that project, cumulative refer- ence intervals of dozens of amino acids and acylcarnitines were derived ‘This work has paved the way to the pursuit of precision new-born screening, which is defined as near- Born in Venice, Italy, Dr Piero Rinaldo is a Professor of Laboratory Medicine and the T. Denny Sanford Professor of Paediatrics. He received his medical and research training at the University of Padova, in Italy (1977-1987) and Yale University in New Haven, Connecticut (1987-1998). A paediatrician, board- certified in Clinical Biochemical Genetics, he co-directs the Biochemical Genetics Laboratory and is vice-chair of Information Management in the Laboratory Medicine and Pathology dept at Mayo Clinic, Rochester, Minnesota. zero FPR preceding any confirmatory use of molecular testing,’ Rinaldo added. The software offers a number of benefits for clinicians. ‘In the context of newborn screening, a substantial reduction of false positive outcomes means better utilisation of health professional time and resources (no need to see patients who don’t need to be seen), and prevention of unnecessary follow up testing.’ Significantly, it reduces unneces- sary stress to parents that something ‘may’ be wrong with their baby when the screening can clarify this.

D I G I TA L PAT H O LO G Y 2 7 Bridging the gap between pathologist and algorithm A human will always be in the loop Digital pathology can gain huge benefits from rapid image search and the effective extraction of knowledge from large medical archives via artificial intelligence (AI). delivered by Dr Hamid Tizhoosh, at the Digital Pathology and AI Congress in London this December. Despite shortcomings, could AI replace the pathologist? Tizhoosh, from the University of Waterloo in Ontario, Canada, and director of the Knowledge Inference in Medical Image Analysis (KIMIA) Lab, outlined how AI uses software algorithms to represent data, classify data, and search for similar instanc- replace the pathologist. ‘The answer to that is no,’ Tizhoosh confirmed. ‘AI is not at the stage where it can write a pathology report but you can use AI to search and find simi- lar cases so that the pathologist can write a better report. ‘The human operator is the ulti- mate validation of AI, so that means there will be no full automation and the human will always be in the loop from the beginning – either to teach, validate, accept or reject. It facilitates identification of anatom- ical and pathological similarities, significantly enhances the clinical workflow, and ultimately paves the way for more informed diagnosis and better patient outcome. Effective archive searching bridg- es the gap between computer algo- rithms and pathologists, adding a new dimension to clinical practice in an ever-more demanding working environment. Learning to search in large image archives The subject was explored in the keynote presentation ‘Learning to search in large image archives: how AI agents can serve pathologists’ Dr Norbert Nowotny is a professor at the Institute of Virology at the School of Veterinary Medicine in Vienna and at the College of Medicine at the Mohammed Bin Rashid University of Medicine and Health Sciences in Dubai (UAE). The virologist’s research focuses on different aspects of (primarily viral) infections of human beings, farm animals, pets, zoo animals and wild animals. Other research interests are medical and veterinary medical entomology, viral infections transmitted by mosquitos and ticks, newly identified infection diseases and zoonoses. He served as local scientific organiser and member of the scientific committee of IMED 2018 and all six previous IMED conventions. es, either in a supervised way (with a pathologist’s direct and indirect instructions) or non-supervised. ‘Learning is just distance minimi- sation between what AI is calculat- ing and what pathologists have put forward as labelled or annotated data,’ he explained. The deep learning aspect of AI has proved most effective in this respect with the ability to classify images (such as deciding whether an image is cancerous or not, or adding grading); segment by focus- ing on a specific part of the image; or search the archive to unearth similar cases. ‘Image search is the unsupervised part,’ he said, ‘because you do not need a pathologist to teach the soft- ware about what to do.’ Whilst he highlighted AI short- comings, he also posed the ques- tion as to whether AI can ultimately ‘You cannot replace the patholo- gist because AI is still very weak and it is the pathologist who writes the detailed report, whereas AI just gives a simple answer. Algorithms cannot provide explanations such as why it came to a decision, or why do you think the image depicts cancerous regions. ‘Artificial intelligence application is just to assist. So, asking whether AI can replace the pathologist is not a serious question.’ AI must be able to multi-task and have human level intelligence To truly have AI that can perform at this level will require it to be capable of multi-tasking and to have human level intelligence such as ethical cognition, he pointed out. However, the existing benefits are significant with AI able to rapidly p e t S y k c u L / k c o t s r e t t u h S Hamid Tizhoosh is director of the Knowledge Inference in Medical Image Analysis (KIMIA) Lab in the Engineering Faculty at the University of Waterloo. He is also a member of the Waterloo AI Institute, and a faculty affiliate to the Vector Institute. As part of his commercial activities, he is presently the AI advisor of Huron Digital Pathology, St Jacobs, Canada. search out similar cases with anno- tated information. A big advantage of this is the element of ‘virtual peer review’ because many of the reports in the archive will be from different pathologists. Needed: around a half million images ‘AI can access the knowledge of colleagues even when they are not there, but you need a large archive of images for that, let’s say, upwards of half a million,’ Tizhoosh said. For an image search in such large archives to be effective, bar-coding (binary tagging) is the key step. ‘You can search half a million scans a second,’ he continued. ‘What you can’t do is bring up 500 glass slides from the basement and search through them – this is where we really start to think about the advan- tages of digital pathology and if you have a large enough archive the search can really contribute to con- sensus.’ (MN) Marilyn Bui is Professor of Pathology and President of Medical Staff and Director of Analytic Microscopy Core, Moffitt Cancer Center, Tampa, Florida. She is chair of the College of American Pathologists (CAP) Guidelines Committee Expert Panel for Qualitative Image Analysis of HER2 immunohistochemistry for Breast Cancer, also vice-chair of the CAP Digital Pathology Committee and President-elect of the Digital Pathology Association and editorial board member of the Journal of Pathology Informatics. is a willingness among health- care professionals and industry to work together to improve the way healthcare computer systems share information and create improved interoperability for digital pathology imaging. The culture among practitioners, however, remains a hurdle. Bui sug- gested a way to overcome this by embracing a planned and phased adoption strategy and to be willing to learn from early adopters. Nonetheless, there are many opportunities offered by digital pathology and AI. ‘In the area of precision medi- cine, the role of pathologists has been increased because now we have augmented ability to produce impactful data for the clinical team. ‘Finally,’ she concluded, ‘digital pathology and AI are here to stay and will continuously transform the delivery of precision medicine. Collaboration of pathologists, scien- tists and industry are important to move the field forward in a mean- ingful way and each individual can make a difference and be a catalyst for change.’ (MN) two golden rules: ‘Don’t get too close to young dromedaries and if you ride on an adult dromedary wash your hands afterwards with any commercially available disinfect- ant.’ It was also suggested to vacci- nate young dromedaries in order to break the infection cycle. However, there are questions on MERS-CoV that remain to be answered: Dromedaries are infected not only on the Arabian Peninsula but also in Africa and Iran – so why have all human cases reported so far originated on the Arabian Peninsula? One theory is that there are regional genetic differences. Moreover, diag- nosing the disease is not always easy, and on the Arabian Peninsula dromedaries are pets, rather than farm animals as in Africa – which means closer contact. But even the infection path among dromedaries is far from clear as Nowotny explained. ‘Is the virus limited to dromedaries or are other animals involved, such as ticks or small rodents?’ These questions the Austrian virologist is currently work- ing to answer. www.healthcare-in-europe.com Ki67, ER, PR, CD3 Imaging•Fully automatic•For WSI or microscope camera•LIMS IntegratedDigital Pathology Solutionswww.vmscope.de