Radiology AI Podcast | RSNA Episode 62: Hot Takes: Radiology Reimagined with AI
Mar 12, 2026
Dr. Dania Daye, associate professor and vice chair for practice transformation who applies machine learning to interventional radiology, discusses how AI is changing radiology practice. She covers governance and who should decide deployments. She talks about measuring real clinical value versus simple efficiency gains. She outlines monitoring for model drift, training clinicians to spot failures, and which human roles remain essential.
AI Snips
Chapters
Transcript
Episode notes
Radiologists Must Steward AI Outputs
- AI is transforming radiology from pure image interpretation to managing AI outputs integrated across workflows.
- Dania Daye emphasizes radiologists remain responsible for final reports and must steward AI outputs rather than act only as product managers.
Include Radiologists In AI Governance
- Build institutional AI governance that explicitly includes end users in deployment decisions.
- Dania Daye points to a published governance framework recommending radiologist representation to avoid disconnects between leadership and clinical usefulness.
Outcome Metrics Outweigh Speed Metrics
- Common AI success metrics focus on process measures like speed and throughput, not clinical utility.
- Daye argues true AI value lies in outcomes, margins, and access, which are harder to measure but crucial for ROI.