From uncertainty to clarity: AI support in difficult cases

In this clinically grounded reading session, Dr. Sophia Blum presents difficult real‑world cases to demonstrate how artificial intelligence can support decision‑making in mammography and digital breast tomosynthesis. Drawing on extensive experience from a university breast center, she highlights where AI meaningfully enhances detection, prioritization, and workflow – and where radiologist expertise remains essential.

The session uses case narratives to illustrate the practical strengths and limitations of AI, including scenarios involving subtle distortions, asymmetric densities, and complex backgrounds. Patient acceptance and trust are briefly addressed to complete a balanced picture of responsible, human‑centered AI integration.

Cases cover difficult to interpret findings: subtle abnormalities where AI may influence prioritization or decision‑making, including

• subtle or easily overlooked features: like small architectural distortions, mild asymmetries, or low‑contrast lesions that benefit from AI support.
• variability in breast densities where AI’s strengths and limitations become more apparent.
• divergent AI and human impressions where AI correctly flags an abnormality the reader might under‑prioritize, or where AI is misleading but requires expert interpretation.
• real workflow challenges like postoperative changes or ambiguous prior comparisons, where AI input influences but not determines the final decision.