A United States study evaluated ProFound AI from iCAD, an AI tool designed to enhance breast cancer detection in digital breast tomosynthesis (DBT) screening. The study analyzed 16,729 DBT exams, including 10,322 before and 6,407 after AI implementation, across three community sites in Indiana. ProFound AI functioned as a concurrent reader, highlighting suspicious regions and providing lesion and case scores to radiologists.

AI integration led to significant performance gains. The cancer detection rate rose from 3.7 to 6.1 per 1,000 exams (P = .008), while the abnormal interpretation rate declined from 8.2% to 6.5% (P < .001). Positive predictive value 1 (PPV1), the proportion of positive screening exams that were true cancers, rose from 4.2% to 8.8%, and positive predictive value 3 (PPV3), the proportion of biopsied cases that were malignant, increased from 32.3% to 56.5% (both significant), reflecting fewer unnecessary biopsies and a higher diagnostic yield. Overall, specificity improved significantly to 94.0% (from 92.1%), while sensitivity showed a modest, nonsignificant increase to 88.4% (from 85.1%).

These findings provide promising evidence that AI-supported DBT screening can substantially enhance cancer detection, diagnostic yield, and specificity while reducing false-positive recalls. As the authors state, although this single-institution study with a small radiologist sample and a predominantly White population limits generalizability, larger and more diverse real-world studies are essential to fully confirm the efficacy of AI breast cancer detection systems.

Improved Breast Cancer Detection with Artificial Intelligence in a Real-World Digital Breast Tomosynthesis Screening Program

Clinical Breast Cancer, 2025

Abstract

Objective

The purpose of this study is to compare radiologists’ breast cancer screening performance before and after the implementation of an artificial intelligence (AI) detection system for digital breast tomosynthesis (DBT).

Materials and Methods

This retrospective study included 4 radiologists reading DBT screening mammograms across 3 clinical sites during 2 distinct time periods. The pre-AI time period from September 1, 2018 to August 31, 2019 included 10,322 standard DBT interpretations with a computer-aided detection system. The post-AI from January 1 to March 18, 2020 and May 4 to December 31, 2020 included 6,407 DBT interpretations with concurrent use of a deep learning AI support system. Endpoints included cancer detection rate (CDR), abnormal interpretation rate (AIR), and positive predictive values for cancer among screenings with abnormal interpretation (PPV1) and biopsies performed (PPV3). Estimates and 95% confidence intervals (CIs) for each radiologist were calculated for each time point and the difference across time periods.

Results

The CDR per 1000 exams increased from 3.7 without AI to 6.1 with AI (difference 2.4, P = .008, 95% CI: 0.6, 4.2). The AIR was 8.2% without AI and 6.5% with AI (difference –1.7, P < .001, 95% CI: –2.5, –0.8). The PPV1 increased from 4.2% to 8.8% with AI implementation (difference 4.6, P < .001, 95% CI: 3.0, 6.3) and PPV3 increased from 32.3% to 56.5% with AI support (difference 24.2, P = .033, 95% CI: 2.0, 46.4).

Conclusion

Real-world interpretation of DBT after implementation of an AI detection system resulted in increased CDR, reduced AIR, and significantly increased PPV1 and PPV3.