Radiology AI Reality Check: Automated Reports, Implementation Failures and Agentic Future

Episode notes

In this episode of Frame by Frame: Rethink Imaging, host Chris St John welcomes Dr. Woojin Kim, an expert at the intersection of radiology and AI innovation, to discuss the future of generative AI in medical imaging. As Chief Strategy Officer at HOPPR and Chief Medical Officer at the ACR Data Science Institute, Dr. Kim draws from deep experience in both clinical practice and healthcare entrepreneurship. Together, they explore the growing capabilities of AI in radiology, from automated draft reporting for chest X-rays and CT scans to the emerging potential of Agentic AI as a true assistant within clinical workflows. Dr. Kim emphasizes that while AI holds the promise of boosting efficiency, with some studies reporting 15.5% gains, it must be grounded in personalization and supported by human oversight to avoid risks like hallucinated results. The conversation dives into practical mechanisms for safe integration, including Model Context Protocols (MCPs) that embed clinical nuance into AI outputs, and the need for hands-on experience over purely theoretical exposure. Dr.