Studies Highlight AI’s Role in Detecting Aggressive Breast Cancers and Supporting Radiologists

At the European Society of Breast Imaging (EUSOBI) Annual Scientific Meeting in Aberdeen, two major studies are drawing attention to how artificial intelligence may reshape breast cancer screening. Both investigations evaluated Hologic’s Genius AI® Detection 2.0 system, with findings that suggest AI can not only match human expertise in interpreting mammograms but may also help identify tumors with more aggressive features.

The first study, conducted at Massachusetts General Hospital, looked at whether AI scores assigned to digital breast tomosynthesis exams reflected tumor biology. Researchers reviewed roughly 600 biopsy-confirmed cancers from exams performed between 2016 and 2019. Most were invasive cancers (80%), with the remainder ductal carcinoma in situ. Hologic’s system assigns a “case score” based on the likelihood that suspicious findings indicate malignancy, ranging from lower scores that correspond to modest risk to higher scores strongly linked to cancer.

Analysis revealed that higher AI case scores correlated with tumors of higher histologic grade and node-positive status — both features of more aggressive disease. According to lead presenter Dr. Manisha Bahl, Associate Medical Director of Quality at Mass General Brigham and Associate Professor of Radiology at Harvard Medical School, this finding points to a potential role for AI in triaging patients for earlier intervention. “Our study suggests that AI may help flag the cancers that matter most, the ones more likely to progress quickly,” Bahl noted. “Detecting these cases sooner could allow treatment to begin earlier, which is often critical for patient outcomes.”

A second study, led by Professor Yan Chen of the University of Nottingham, tested Genius AI Detection in a direct comparison with radiologists. The project involved 108 radiologists from both the U.S. and U.K., each asked to interpret 75 challenging breast cancer cases. The AI system analyzed the same set. Early results indicate that AI achieved performance comparable to that of the radiologists, with a tendency toward higher sensitivity but lower specificity.

This balance, researchers noted, suggests that AI could serve as an additional safety net, especially in screening programs where two radiologists are expected to review each mammogram. In regions experiencing workforce shortages, AI could help ensure double-reading standards are maintained without overburdening radiology teams.

Hologic executives emphasized that the findings demonstrate the promise of AI to enhance accuracy and efficiency in screening while offering new insights into tumor characteristics. “These results underscore AI’s ability to make breast cancer screening more effective and more personalized,” said Mark Horvath, President of Breast and Skeletal Health Solutions at Hologic. “We’re committed to advancing this technology to support clinicians and improve care for women worldwide.”

In addition to its AI initiatives, Hologic used the EUSOBI congress to highlight new surgical tools, including the Sentimag® Gen 3 system for breast tumor localization and staging, developed by its U.K.-based subsidiary Endomag. Educational workshops and training sessions accompanied the presentations, signaling the company’s push to integrate innovation across the full spectrum of breast cancer care.

Together, these studies provide evidence that AI can meaningfully assist radiologists — not as a replacement, but as a complement — while also helping to prioritize the cancers where early detection can have the greatest clinical impact.