AI breast density assessment shows low sensitivity for dense tissue in women with Gilbert's Syndrome

A CNN for breast density assessment in postmenopausal women with Gilbert's Syndrome showed only 55.6% sensitivity for dense breasts, though specificity reached 100.0%. Findings are preliminary due to a very small sample (n=21) and warrant cautious interpretation in this specific…

• Prospective cross-sectional pilot study of 21 postmenopausal women, comparing a convolutional neural network (CNN) against radiologist breast density assessments.
• In the Gilbert's Syndrome group, kappa agreement for dense breasts was fair (0.381) vs good (0.700) in wild-type controls.
• Primary limitation is the extremely small sample size (n=21), resulting in wide confidence intervals and limiting external validity.

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