Two papers were published this week, one in Radiology that compared the performance of radiologists with an AI algorithm in reading mammograms, and the other in Academic Radiology that employed a deep-learning AI model to classify breast ultrasound images as benign or malignant. Dr Liane Philpotts, a prominent breast imaging specialist at Yale (right), commented, “As a second reader, it appears AI has a definite role that should ease the demanding job of reading large volumes of screening mammograms.” However, is she perhaps ignoring the elephant in the room: Will AI make radiologists redundant?
This was the provocative heading of an article published last week in Financial Mail, one of South Africa’s leading business magazines. The author based his article on interviews with local academics and clinicians. Professor Athol Kent, an ObGyn specialist affiliated with the University of Cape Town (UCT), who runs a Journal Article Summary Service, observed, “AI will move us forward … and we the doctors will have to figure out how AI works to our advantage and stay informed.”
Dr Deshen Moodley, an associate professor of computer science at UCT (left), was more outspoken, warning that for medical professionals who are resistant to AI, “The train has already left the station. Physicians will have to scramble to catch up. If they don’t adjust, they’ll be left behind.” Given the healthcare challenges that South Africa faces, Moodley is nevertheless optimistic that AI could have a significant impact so long as there are open-source platforms that will reduce the cost of delivery.
Dr David Jankelow, a cardiologist in private practice in Johannesburg (below right), is currently collaborating with clinicians at the Mayo Clinic in the USA. He has been impressed with the ability of an AI algorithm to combine ECG readings and ultrasound images to assess heart function. He said, “If you’re a radiologist, you’re already over the edge of the cliff – you just haven’t looked down yet. There’s no ground underneath. In five years, deep learning is going to do the job better than radiologists. They should stop training them now.”
Coming from a cardiologist, most radiologists would consider such a statement the height of cheek! We previously highlighted the venture capitalist Vinod Khosla who in June 2018 predicted that radiologists would be obsolete within five years because of sophisticated AI algorithms. Well, that hasn’t come to pass although we should not ignore the concerns that these arguments might make it more difficult to recruit medical students to pursue a career in radiology. Rather than making radiologists redundant, AI should be seen as a friend, not a foe.