Radiology AI Podcast | RSNA

Satvik Tripathi sits down with Saurabh Jha for a candid conversation on where radiology AI is delivering real value versus hype. They explore automation, workflow impact, and the hard questions shaping the future of AI in radiology. Radiology: Artificial Intelligence

Host Ali Tejani speaks with Dr. Laurens Topff and Stephane Willaert about how to identify high-value clinical problems in radiology and translate AI tools from research prototypes into real clinical practice. Together, they discuss lessons learned from collaboration, workflow integration, and what it takes to develop AI that radiologists will actually use. https://pubs.rsna.org/journal/ai

Our hosts, Ali and Paul, speak with Dr. Udunna Anazodo and Dr. Marouf Adewole about their groundbreaking work on the BRATS Africa challenge and building AI-ready brain tumor imaging datasets across Nigeria. They share insights into the challenges of medical imaging in resource-limited settings, the power of global collaboration, and how their efforts are shaping the future of inclusive AI in radiology.

In this episode of the Radiology Artificial Intelligence Podcast, host Dr. Paul Yi speaks with Drs. Eric and Kevin Wu, recent Stanford PhDs, about their journey through academia, industry, and the startup world. They dive into their latest project, MedArena, a physician-powered platform designed to evaluate medical LLMs, and explore how AI can be more effectively integrated into real-world clinical workflows.

In this AI-generated episode of Radiology AI Papers in a Capsule, we discuss a study that extends the NeuroHarmony AI model to address scanner variability in brain MRI for Alzheimer's disease assessment. Learn how incorporating cognitive status into harmonization may improve the reliability of quantitative imaging across diverse clinical settings. A Machine Learning Model to Harmonize Volumetric BrainMRI Data for Quantitative Neuroradiologic Assessment ofAlzheimer Disease. Archetti and Venkatraghavan et al. Radiology: Artificial Intelligence 2025; 7(1):e240030.