A recent study published in the American Journal of Roentgenology compares false positives in digital breast tomosynthesis (DBT) exams interpreted by artificial intelligence (AI) versus those evaluated by radiologists. Analyzing data from nearly 3,000 women, the study found that while both AI and radiologists reported a 10% false positive rate, AI flagged nearly three times the number of false positives (932) compared to radiologists (315). Additionally, AI-only assessments were associated with fewer cases of dense breast tissue, indicating potential advantages in distinguishing between dense tissue and actual abnormalities. The overlap in false positives between AI and radiologists was noted to be 13%, with a significant percentage of concordant findings leading to biopsies that resulted in high-risk lesions. Researchers suggested that focusing on these concordant findings might reduce unnecessary recalls while maintaining diagnostic effectiveness. The study also acknowledged its limitations, including reliance on a single AI tool and mammography vendor.
Fri, 03 Oct 2025 10:34:57 GMT | diagnosticimaging.com