Radiology AI Podcast | RSNA Episode 60: Hot Takes: Rapid Questions on the Future of Radiology AI
Dec 22, 2025
Saurabh Jha, associate professor and candid radiology AI commentator known as Rogue Rad, cuts through hype with blunt takes. He contrasts vendor vs homegrown tools, predicts automation wins like chest x‑ray measurements, warns about risky copilot UIs and legal traps, and highlights radiomics, scan acceleration, and equity challenges in adoption.
AI Snips
Chapters
Transcript
Episode notes
Vendor Hype Outpaces Hospital Adoption
- Vendors drive much of current AI visibility while hospitals report less adoption than vendor claims.
- Foundation models look promising but have not yet proven workflow efficacy.
Automate Measurements, Then Integrate
- Automate routine image measurements that are easily segmented, like vascular dimensions and ejection fraction.
- Integrate measurements into the report and radiology information system to deliver real workflow value.
Choose Homegrown AI With Clear Purpose
- Build an internal AI team when you want lasting ROI, scientific progress, and control over clinical use-cases.
- Ensure a strong use case, clear purpose, and align development to implementation rather than pet research projects.