0
0
Lund, Sweden – A new report from Lund University’s groundbreaking MASAI trial reveals that AI-supported breast cancer screening is now 29% more accurate in detecting cancer compared to traditional methods. This significant leap in performance highlights AI’s potential to revolutionize early detection, particularly for invasive cancers, offering better prognosis and more effective treatment options for patients.
The updated findings, recently published in The Lancet Digital Health, show a marked improvement over last year’s results. The MASAI trial, a randomized study initiated in 2021, evaluates the impact of AI on mammography screening. The second phase of the trial reveals AI’s ability to detect not only more cancers but also a higher proportion of early-stage and more aggressive cases, which are critical to detect for improved survival rates and reduced treatment costs.
Kristina Lång, a researcher and associate professor of diagnostic radiology at Lund University, and the lead author of the study, expressed her excitement over the improved results. “Since our first report, the number of cancers detected by AI-supported screening has increased by 9%, now detecting 29% more cancers than traditional screening,” she said. “AI is especially effective in identifying early-stage invasive cancers, including aggressive forms that are crucial to catch early.”
The trial involved nearly 106,000 women who were randomly assigned to either traditional mammography screening or AI-supported screening. The results showed AI detection was more accurate overall, identifying 338 cases of cancer compared to 262 found by traditional screening. Additionally, AI-supported screening detected 24% more early-stage invasive cancers and 51% more in situ (pre-cancerous) cases.
The potential for reduced suffering, higher survival rates, and lower healthcare costs is significant, with AI’s ability to find more cancers early leading to fewer complications and less intensive treatment later on. Importantly, despite the increase in cancer detections, the AI screening did not significantly increase false positives, which is a critical concern for women’s mental health.
Lång also emphasized AI’s ability to reduce radiologists’ workload by 44%, helping to address the current shortage of breast radiologists in Sweden. AI support in screening has already been integrated into several regional programs, improving efficiency and accessibility.
While the MASAI trial has shown promising results, researchers are now focusing on “interval cancers”—cases that develop between routine screenings. Lång and her team are hopeful that AI will also prove beneficial in reducing the number of missed cancers in these intervals.
The final report of the MASAI trial will be published next year. For now, the trial’s results continue to demonstrate AI’s immense potential to improve breast cancer screening accuracy and efficiency, and its ongoing impact is already shaping Sweden’s national screening programs.
Disclaimer: The results presented are part of an ongoing research study, and while the findings are promising, further studies are required to confirm the long-term benefits and scalability of AI-assisted mammography in breast cancer detection.
For further reading, the study titled “Screening performance and characteristics of breast cancer detected in the Mammography Screening with Artificial Intelligence trial (MASAI)” is available in The Lancet Digital Health (2025). DOI: 10.1016/S2589-7500(24)00267-X.
