Research continues to advance methods for the early detection of breast cancer, aiming to identify tumors that may not be palpable or accompanied by symptoms, and to improve the quality of examinations used in screening programs.
One important recent study in this field is the MASAI trial, published in The Lancet on January 31, 2026, which compared AI-supported mammogram reading with conventional double reading by two radiologists.
How was the study conducted?
This randomized controlled trial, conducted in Sweden, included 105,934 women who were assigned to two groups. Artificial intelligence was used to support radiologists and guide image reading, with clinical assessment remaining part of the process.
What were the main findings?
Sensitivity for cancer detection increased from 73.8% to 80.5%. Sensitivity refers to a test’s ability to detect cancer when it is present.
Specificity remained at 98.5% in both groups, meaning that the improvement was not accompanied by a decline in the ability to correctly classify women without cancer.
The rate of cancers diagnosed between screening rounds was 1.55 versus 1.76 per 1,000 women. The new approach met the criterion for non-inferiority, but the difference did not establish statistically significant superiority in reducing these cancers.
These findings support improvements in the quality of mammogram reading. However, they do not demonstrate a reduction in mortality, nor do they compare screening with no screening. Therefore, they should not be presented as standalone evidence quantifying the benefit of screening.
The International Health Organization (IHO) emphasizes the importance of responsibly applying scientific advances within a care pathway that connects detection with diagnosis, follow-up, and treatment.
Scientific reference:
Gommers J, et al. The Lancet. 2026;407(10527):505–514.
DOI: 10.1016/S0140-6736(25)02464-X
