Using artificial intelligence in mammography tests “is as good as two breast radiologists working together to detect breast cancer”, according to new research.
The study, published in the Lancet Oncology journal, involved more than 80,000 Swedish women and the results almost halved the screen-reading workload for radiologists without increasing false positives.
However, the researchers noted that “primary outcome results” are not expected for several years and these will establish whether AI reduces interval cancers (those diagnosed between screenings) in 100,000 women with at least two years of follow-up – and ultimately whether AI’s use in mammography screening is justified.
The findings were described as an “interim safety analysis of the first randomised controlled trial of its kind using AI, and underscore AI’s potential to make mammography screening more accurate and efficient”.
‘Godfather of AI’ Geoffrey Hinton warns of ‘quite scary’ dangers of chatbots as he quits Google
A super-charged technology threatens to destroy the integrity of the Leaving Cert
Big tech has been stealthily training its AI models. Creatives are finally waking up to the dangers. Are they too late?
Mark O’Connell: Had this technology existed when I was a student, I might not have been above using it
[ Scientists use AI to find new antibiotic to treat deadly superbugOpens in new window ]
AI-supported screening detected 20 per cent more cancers compared with the routine double reading of mammograms by two breast radiologists.
“These promising interim safety results should be used to inform new trials and programme-based evaluations to address the pronounced radiologist shortage in many countries,” said lead author Dr Kristina Lång of Lund University. “But they are not enough on their own to confirm that AI is ready to be implemented in mammography screening.
“We still need to understand the implications on patients’ outcomes, especially whether combining radiologists’ expertise with AI can help detect interval cancers that are often missed by traditional screening, as well as the cost-effectiveness of the technology.”
Breast cancer screening with mammography improves prognosis and reduces mortality by detecting breast cancer at an earlier and more treatable stage. However, estimates suggest that 20 to 30 per cent of “interval cancers” that should have been spotted at the preceding screening mammogram are missed – while suspicious findings often turn out to be benign.
The research was welcomed by Irish radiologists specialising in breast screening.
Prof Fidelma Flanagan, lead adviser to BreastCheck and clinical director of its Eccles Screening Unit attached to the Mater Hospital in Dublin, said it indicates the very promising prospect of improving breast cancer screening and “confirms that there definitely will be a role for AI”.
While it suggests AI is as good as a radiologist in screening, she said work needed to be done in real-life scenarios. “We will get there ... AI will complement radiography but it won’t take over,” Prof Flanagan said.
[ Hello, my name is Dr ChatGPT: Are robot doctors really the future?Opens in new window ]
There is no population screening for breast cancer using AI at present. In effect, she said, AI will “add to the workforce”, allowing radiologists to increase their interactions with patients.
AI may enable an extension of the age range in screening programmes and facilitate increasing checks of breast density – an important emerging priority. It would also allow facilitate work on more complex cases and emerging technologies, Prof Flanagan said.
“It is not ready yet to transfer into population screening but studies like this one along with prospective trials are necessary if this transition is to happen,” she said.
BreastCheck are looking forward to incorporating AI into the Irish breast screening programme once it is safe and ready, she confirmed. It could potentially assist with increasing the screening volumes, such as extending the current age range. Also by freeing up radiology reading time it could allow for more direct patient interactions and assessments.
“AI may also have a role in helping intensify screening along with supplemental imaging in higher risk sub population groups such as family history cohorts or women with dense breast tissue.”
Ireland has limited radiology resources, which was highlighted after the Covid-19 pandemic restrictions created backlog. “We have to look outside the box and add resources. This [AI] can be hugely valuable.”
Innovation and investment in cancer means that across the areas of prevention, early detection and treatment, the future is promising, said Dr Claire Kilty, head of research with the Irish Cancer Society.
“Artificial intelligence has the potential to enhance what we know about someone’s risk of cancer as well as the biology of a tumour should a cancer be found. However, there are still questions to ask about the extent to which AI can enhance the detection rate of radiologists, the acceptance amongst the public of AI as a tool in screening, as well as where on the screening pathway AI is best used,” she added.
While initial findings showing AI has the potential to be safely used to improve mammography and reduce workload burden on radiologists, Dr Kilty said, “there is still a long way to go before conclusive findings are published and we will be watching developments with interest while continuing to advocate for measures that are in the best interests of patients.”
European guidelines recommend double reading of screening mammograms by two radiologists to ensure high sensitivity and to correctly identify those with disease. But there is a shortage of breast radiologists in many countries, and it takes more than a decade to train a radiologist capable of interpreting mammograms.
Overall, AI-supported screening resulted in a cancer detection rate of six per 1,000 screened women, compared with five per 1,000 for standard double reading without AI.
