AI shifts BI-RADS calls and boosts confidence among radiology residents reading mammograms

In 112 mammograms read by 10 residents, AI changed the BI-RADS category in 9.7% of assessments and altered diagnostic confidence in 19.2%, with upgrades outnumbering downgrades. Sensitivity for BI-RADS 4+ rose from 72.0% to 82.3% against an expert-consensus standard, though this…

• Retrospective, single-center, paired multi-reader study: 10 residents read 112 exams (2230 resident-breast interpretations) first unassisted, then with AI output.
• AI-associated modifications occurred in 24.3% of reads; confidence more often increased than decreased; agreement with the expert BI-RADS consensus improved only modestly.
• Limitation: reference standard was expert consensus, not histopathology, so the observed sensitivity gains do not directly translate to confirmed cancer-detection improvement.

RadPigeon summaries are original and for information only. They are not clinical advice.