Vicky Goh
Professor of Cancer Imaging at King’s College London and Guy’s and St Thomas Hospitals
Artificial intelligence and big data approaches could tackle imaging backlogs and aid personalised medicine by making science fiction a reality.
Imaging backlogs could be aided by artificial intelligence
The increasing growth in demand for imaging services and the backlog in reporting cannot be tackled by simply recruiting more radiologists. Artificial intelligence (AI) and big data approaches could help, with technological solutions that are moving from science fiction into reality.
“It is time to take a different approach in Radiology. AI is one approach that could be used to aid doctors in diagnosis,” says Vicky Goh, Professor of Cancer Imaging at King’s College London and Guy’s and St Thomas Hospitals.
It could be done – in fact it is on its way. Goh says: “Computers can be trained to recognise abnormal scans. By collecting data from a huge number of images which have already been reported, and feeding them into a computer, it can ‘learn’ to distinguish between the normal and the abnormal.”
Machines are able to assist in diagnosing some cancers
This is happening now. Giovanni Montana, Professor of Bioinformatics at Kings College London, is using a million Chest X-Rays and reports, plus a sophisticated algorithm, to train a computer to differentiate normal and abnormal images. The work is ongoing so the technology is not in clinical use yet.
However computer-aided diagnosis is already possible in a limited way in screening programmes of breast cancer. “Machines will never replace doctors but algorithms can be very good at detection, so they can be used in detecting cancers,” Goh says. “However, it is a big step from detection to fully automated accurate diagnosis.” This would require far more sophisticated technology.
Goh says: “Imaging is so sophisticated now. In cancer we can already collect information about the size, shape, texture of cancers, its physiology, and its metabolic activity with molecular imaging. In the future we can envisage an integrated system that could hold all of the medical information about a patient, including family history, test results including genetics, and radiology images and reports. By combining all of that information, we could train machines to predict the likely diagnosis and maybe even outcome.”
The future of imaging looks more like science-fiction
This technology is not here yet, but it is possible – and it could potentially enable highly-personalised medicine. “It will be some time before the necessary technology is available – and affordable – but we are progressing towards it in incremental steps,” says Goh.
It may not be long before the technology of science fiction becomes a reality. Dr ‘Bones’ McCoy, the doctor on the Star Ship Enterprise in Star Trek, used a hand-held device which he swept across the patient’s body to diagnose medical problems.
Goh says: “That kind of diagnostic scanning could be the future, perhaps not in my lifetime, but in the lifetime of my children.”
