AI-supported breast tumour screening found more cancers, finds study
New Delhi: Mammography screening supported by artificial intelligence (AI) detected one-fifth more breast cancers compared with standard screening and nearly halved the screen-reading workload for radiologists, new research published in The Lancet Oncology journal said.
The findings conveyed the safety of AI-supported mammography, a diagnostic procedure for breast tumour screening, as an alternative to the conventional double reading by radiologists, the study led by Lund University, Sweden, said.
Eighty-thousand Swedish women were recruited for the study and randomly allocated into two groups - an intervention group (40,003 women) that underwent AI-supported screening and a control group (40,030) that underwent standard screening.
In total, 244 women (28 per cent) recalled from AI-supported screening were found to have cancer compared with 203 women (25 per cent) recalled from standard screening - resulting in 41 more cancer detections with the support of AI.
"We found that using AI resulted in the detection of 20 per cent (41) more cancers compared with standard screening, without affecting false positives. A false positive in screening occurs when a woman is recalled but cleared of suspicion of cancer after workup," said Kristina Lang, lead researcher and associate professor in diagnostic radiology at the university.
The false-positive rate was 1.5 per cent in both study arms, the researchers said. Further, the screen-reading workload for radiologists was found to be reduced by 44 per cent. The screen-readings with AI-supported screening were 46,345 compared with 83,231 in standard screening.
Five months of a radiologist's time were estimated to have been saved by using the technology to read the roughly 40,000 screening examinations in the AI group.
Explaining the basis for this estimation, Lang said that, on average, a radiologist read 50 mammograms an hour.
On the other hand, AI failed to provide a risk score in 0.8 per cent of the cases that were referred to standard care (double reading), the study said.
Called the Mammography Screening with Artificial Intelligence (MASAI) trial, the study said it was the first randomised controlled trial evaluating the effects of AI-supported screening.
"In our trial, we used AI to identify screening examinations with a high risk of breast cancer, which underwent double reading by radiologists. The remaining examinations were classified as low risk and were read only by one radiologist," said Lang.
The generalisability of the study results is limited, the researchers said, in that the analysis was conducted at a single centre and was limited to one type of mammography device and one AI system.
"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. Just because a screening method finds more cancers does not necessarily mean it's a better method.
"What's important is to find a method that can identify clinically significant cancers at an early stage. However, this has to be balanced with the harm of false positives and the overdiagnosis of indolent cancers," said Lang.
