AI-Enhanced Mammography Improves Risk Prediction in Breast Cancer Screening

Recent findings presented at the European Congress of Radiology suggest that the incorporation of artificial intelligence (AI) into mammography screenings significantly enhances the ability to predict breast cancer occurrence in subsequent screenings. The study, involving more than 135,000 mammograms from 67,000 women, examined the effectiveness of exam risk scores (ExRS) generated by the AI software Lunit Insight MMG in identifying individuals at risk of developing breast cancer.

The research, which averaged a 777-day follow-up period between mammography screenings, revealed that initial breast cancer risk scoring with AI augmentation increased predictive accuracy by approximately 80 percent in subsequent screenings for women who developed breast cancer. Specifically, among the 451 women diagnosed with breast cancer, there was a stark increase in their ExRS from an initial mean of 15.4 to 73.9 by the next screening.

In contrast, those who did not develop breast cancer exhibited a mean ExRS of 6.7 initially and 6.4 upon subsequent examination. This highlights the capability of AI-derived ExRS to differentiate between varying levels of cancer risk starting from the baseline. The consistent performance across different breast density categories (BI-RADS) further underscores its utility, as noted by Dr Clauda Maria Weiss and colleagues from the Treviso Health Authority in Veneto, Italy.

For women with non-dense breasts (BI-RADS categories A and B), the mean ExRS progressed from 13.8 to 73.7 between screenings. In women with dense breast tissue (BI-RADS categories C and D), the ExRS rose from 17.6 to 74.1. These findings suggest that ExRS can serve as a valuable tool for risk-based stratification in breast cancer screenings, offering a more personalized approach to early detection and prevention.

The study emphasizes the potential role of AI in enhancing the precision of mammography screening programs by allowing clinicians to allocate resources more effectively based on individual risk assessments.