Artificial intelligence (AI) is the hot topic these days in a host of industries as well as in medicine and especially radiology. It certainly was the major theme at RSNA last November which opened with an insightful President address focusing on the rise of technologies such as AI and machine learning and their impact on the future of radiology practice. Naturally, ECR couldn’t be outdone on the subject, so AI will be at least as prominent in Vienna this year as it was in Chicago a few months ago. Indeed, interactive exhibits, trainings and a plethora of scientific sessions (a record number of 44 covering 317 presentations) will be devoted to AI at ECR 2019. In addition, the AIX Theatre – located in the heart of the AI Exhibition (AIX) inside the X1 hall – will feature interactive panel discussions and 8-minute pitches from over 20 innovative companies in the field. Thanks to the explosion in the stock of medical images in recent years, combined with the exponential growth in computing power as well as the availability of enormous storage capacities, AI is on the verge of gradually transforming, not just radiology but the entire practice of medicine in the coming decade. This is creating tremendous opportunities for the big vendors (the likes of Siemens, GE and Fujifilm to name but a few) but also for a larger number of smaller technology companies, some of which will enter into partnerships with the big players.
There are two opposing views about how the deployment of AI will affect the radiologist’s job. The conventional opinion is that AI will relieve radiologists from part of the burden – both physical and visual – of reading images of as many as 100 examinations per day (this translates into interpreting a new clinical image every 3 to 4 seconds), a busy radiologist being able to read about 20,000 studies a year. With the number of examinations continuing to grow far more rapidly than the number of radiologists in most countries, radiologists, thanks to AI, will become even more productive as they will no longer need to look at all the images but in many cases (especially for the most routine exams) will simply validate the outcomes provided by the algorithms. The other, more ambitious but potentially controversial view is that AI will radically transform the function of radiologists so that they will in future act as clinicians rather than pure image readers and spend more time in noninterpretive and more patient-centric activities such as interacting with colleagues in other specialties. The question is, are radiologists prepared for this new role?