As reported by Digital Health, artificial intelligence could help detect more breast cancers during screening while easing pressure on radiologists, according to new research
The findings come from two linked studies published on 10 March 2026 in Nature Cancer. Researchers compared the current screening approach, where two clinicians review each mammogram, with a model in which one human reader is paired with an AI system created by Google.
In one analysis, scientists reviewed screening images from about 125,000 women. When AI was used alongside a clinician, more invasive cancers were detected and overall cancer detection improved. The technology also produced fewer false alarms and resulted in fewer women being called back after their initial scan.
The system was also able to identify a quarter of so-called interval cancers. These are cancers discovered between routine screening appointments after a previous mammogram had shown no signs of disease.
A separate part of the study examined 9,266 recent screening cases from two services operating across 12 sites in London. Here, researchers measured how long it took to review scans. Introducing AI shortened reading time by almost one third.
The difference in processing speed was substantial. On average, the AI system produced an assessment in about 17.7 minutes, whereas the first human reader typically took just over two days.
In the UK screening programme, mammograms are normally examined independently by two specialists, usually radiologists. Depending on the system used, the second reader may or may not see the first reader’s decision when making their own assessment.
Be the first to comment