Breast Cancers Detected and Missed by AI-CAD: Results from the AI-STREAM Trial.
Authors
Affiliations (6)
Affiliations (6)
- Department of Radiology, Soonchunhyang University Seoul Hospital, 59 Daesakwan-ro, Yongsan-ku, Seoul 04401, Korea.
- Department of Radiology, Kyung Hee University Hospital at Gangdong, Seoul, Korea.
- Department of Radiology, Nowon Eulgi University Hospital, Seoul, Korea.
- Department of Radiology, Konkuk University Medical Center, Seoul, Korea.
- Department of Radiology, Inje University Busan Paik Hospital, Busan, Korea.
- Department of Radiology, CHA Bundang Medical Center, Seongnam, Korea, Department of Radiology, Yongin Severance Hospital, Yonsei University College of Medicine, Yongin, Korea.
Abstract
Purpose To evaluate the characteristics of breast cancers detected and missed by artificial intelligence-based computer-assisted diagnosis (AI-CAD) during screening mammography. Materials and Methods This retrospective secondary analysis was conducted using data from the Artificial Intelligence for Breast Cancer Screening in Mammography trial (ClinicalTrials.gov: NCT05024591), a prospective, multicenter cohort study performed from 2021 to 2022. AI-CAD results were categorized into nine subgroups based on abnormality scores (in 10% increments). Positive predictive value of recall (PPV1) were calculated for each subgroup and by breast density, and AI-CAD scores were compared with mammographic and pathologic features. Results A total of 24,543 women (mean age, 59.8 years ± 11.2 [SD]), including two with bilateral cancers, were included, with 148 cancers confirmed by a pathology after 1 year of follow-up. AI-CAD results were negative in 23,010 cases (93.8%) and positive in 1,535 (6.2%). The overall PPV1 was 8.7% (133/1,535), with a sensitivity of 89.9% and specificity of 94.3%, PPV1 increased with higher abnormality scores but remained below 3% in groups 1 and 3 for dense breasts. AI-CAD detected 3.4% (5/148) of cancers missed by radiologists but missed 8.1% (12/148) that were detected by radiologist recall. Lower abnormality scores were observed in patients presenting with mammographic asymmetry (<i>P</i> = .001) and luminal A subtype (<i>P</i> = .032). Conclusion AI-CAD shows potential to improve breast cancer detection in screening programs and to support radiologists in mammogram interpretation. Understanding the imaging and pathologic features of cancers detected or missed by AI-CAD may enhance its effective clinical application. ©RSNA, 2025.