Skip to main content
Home » Imaging » What the AI revolution could mean for you
Future of Imaging February 2019

What the AI revolution could mean for you

AI looks set to transform the world of imaging, cutting reading times, boosting accuracy, and improving patient care. Where is it now?


Dr Hugh Harvey

Clinical Director, Kheiron Medical Technologies

Artificial intelligence (AI) has the power to revolutionise imaging, saving time, money and lives.  

Within the next few years it’s forecast to be a standard feature of UK healthcare – but what does this mean for patients, radiologists, radiographers and the wider healthcare team?

Dr Hugh Harvey, a consultant radiologist who sits on the Royal College of Radiologists informatics committee and is an AI advisory board member, says: “Within radiology, I see AI as the future.”

Using AI to help diagnose cancer 

Dr Harvey, who has been using deep learning – a form of AI – to aid breast cancer diagnosis using mammograms, says: “In the past year we have seen the first retrospective studies in the UK, which test the algorithms used in AI systems against the judgement of experienced radiologists in detecting breast cancer. 

“We already have a handful of results, but 2019 will bring more from prospective studies, and within a couple of years we should have a clearer picture of how AI will perform in the real world. The picture looks very promising.”

More than tumour detection  

AI holds out the possibility of faster and more accurate image interpretation, making it easier for radiologists to decide whether cancer is present. But it can do more. Dr Harvey says: “Not only can AI highlight regions of interest on scans, such as nodules in the lungs, bleeds in the brain, and bone fractures, it can assist in radiomics, measuring and analysing findings to uncover otherwise hidden disease characteristics.”

AI can also be used behind the scenes in the radiology department, to help prioritise cases, decide the most appropriate type of scan, reduce the time taken for procedures such as MRIs, cut radiation doses, and help schedule appointments.

It could also aid in recording conclusions and communicating information to clinics and patients.

“No single system can carry out all these tasks yet, but this could be achieved by different systems working together,” says Dr Harvey.

Is AI the end of radiologists?

“No,” says Dr Harvey. “A few years ago, there were overblown fears in the profession that AI meant radiologists would no longer be needed, but now people are realising that algorithms cannot replicate everything radiologists do. We still need human skills to tie together all the information gathered.”

Instead, AI will make life easier, he says. “AI can reduce image reading time and improve accuracy. There is currently a huge image reading backlog and a shortage of radiologists. AI could help close the gaps. That’s why these prospective studies, which aim to show how AI can increase accuracy and save time, are so exciting, though we also need health economics assessments that prove it can pay for itself.”

Wider applications for AI across cancers and chest X-rays

The huge number of images generated by breast cancer screening make it an obvious area for introducing AI, which benefits from having very high numbers of images from which to ‘learn’ to detect abnormalities. However, AI has the potential to revolutionise the detection of a far wider range of conditions.

“Chest X-rays still account for the largest volume of images we generate, and there is already excellent work by research groups going on to design algorithms to analyse them,” says Dr Harvey, “but the range of pathologies revealed by a chest X-ray is far greater than those revealed by a mammogram, so it is a more complicated task.”

We need a central data hub to ‘train’ AI

One of the problems in advancing AI is access to large numbers of images on which to train the algorithms. Dr Harvey has proposed the creation of what he calls the British Radiology Artificial Intelligence Network (BRAIN), a central data science institute to give secure, regulated access to anonymised NHS medical imaging data.

“In France and the US, some medical groups are already collecting data centrally. The UK should be doing this too, as AI is clearly a part of the future of imaging. It’s encouraging to see the creation of new AI ‘centres of excellence’ across the UK, thanks to the Industrial Challenge Strategy Fund”.

February 2019

Next article