AI is better than a radiologist at spotting cancer
A COMPUTER algorithm could be a lifeline for the NHs after proving to be better than a radiologist at detecting breast cancer in mammograms.
trained using X-rays from 91,000 women, artificial intelligence has been found to detect more cases which would have been missed – and reduce wrong diagnoses in healthy women.
the algorithm is not yet better than the NHs – which uses at least two radiologists to check a mammogram – but it could cut the workload of one by almost 90 per cent.
there have been repeated alerts over a shortage of radiologists.
an international team of researchers, including Imperial College London and Google, designed the computer programme and tested it on more than 25,000 breast scans done in the UK between 2012 and 2015.
Compared with a single radiologist, the algorithm missed 2.7 per cent fewer cases of breast cancer. the rate of ‘false positives’, where women are mistakenly diagnosed with breast cancer but do not have it, was cut by 1.2 per cent.
surgeon Professor Lord Darzi of Denham, co-author of a paper on the algorithm from the Institute of Global Health Innovation at Imperial College London, said: ‘screening programmes remain one of the best tools at our disposal for catching cancer early and improving outcomes for patients, but many challenges remain – not least the current volume of images radiologists must review.
‘While the findings are not directly from the clinic, they are very encouraging, and they offer clear insights into how this valuable technology could be used in real life.’ the algorithm was trained using X-rays from 76,000 women in the UK and around 15,000 in the Us. Unlike radiologists, it was not given the health records of the patients but used mammograms alone.
It was able to correctly identify cancers with a similar degree of accuracy to expert radiologists.
the study, published in Nature, then tested the programme on historic scans from 25,856 anonymous women at st George’s Hospital in London and Jarvis Breast screening Centre in surrey.
Women in the UK are invited for breast screening every three years between the age of 50 and 70, with mammograms believed to save around 1,300 lives each year.
Researchers compared the algorithm’s decisions with that of a first radiologist seeing a scan. they assumed images would only be sent to a second person where the computer and radiologist disagreed.
they concluded that the technology could reduce a second radiologist’s workload by 88 per cent.
‘The findings are very encouraging’