Next RISE-MICCAI Journal Club – July 11, 2026
Recent News
Join the next RISE-MICCAI Journal Club Session:
Paper: Making AI models and datasets more transparent
Author: Ricardo Gonzales, Harvard Medical School
Saturday, July 11, 2026 at 12:00 pm EDT / 4:00 pm UTC
Abstract:
Medical AI evidence is increasingly distributed across papers, supplements, code repositories, dataset pages, challenge websites, model cards, and metric tables, making it difficult to find, compare, reproduce, and evaluate models and datasets consistently. This talk presents the Radiology Ontology of AI Datasets, Models and Projects (ROADMAP) and the RSNA Annotated Library of AI Systems (ATLAS) as complementary infrastructure for turning fragmented medical AI documentation into structured, searchable, and reusable evidence. ROADMAP provides a controlled, machine-interpretable vocabulary for describing AI models, datasets, projects, performance metrics, intended uses, users, limitations, data provenance, and evaluation context, while ATLAS operationalizes this framework through public model and dataset cards that can be reviewed, indexed, compared, and queried. I will discuss how this structured evidence layer can support model and dataset discovery, reporting completeness checks, bias-relevant metadata inspection, metric harmonization, regulatory and editorial review, and scalable card creation through large language model pre-fill followed by human curation. The broader goal is to move medical AI from isolated model claims toward transparent, inspectable, and reusable evidence that better supports trustworthy research and clinical translation.
Papers: https://doi.org/10.1148/ryai.260070 and https://doi.org/10.1148/ryai.260069
About the author: Ricardo Gonzales
Ricardo A. Gonzales, DPhil, FSCMR, is a postdoctoral researcher at Harvard Medical School and Massachusetts General Hospital. He develops trustworthy medical-imaging systems that translate complex biomedical data into reproducible biomarkers, analysis tools, and reusable scientific infrastructure. He earned a DPhil in Medical Sciences from the University of Oxford as a Clarendon Scholar and a degree in Electrical Engineering from the University of Engineering and Technology (UTEC) in Peru. His work spans deep learning, image registration, segmentation, quality control, model evaluation, cardiac and brain imaging, and clinical AI implementation, with methods deployed in medical-imaging software and large-scale research workflows. He is also a medical AI data standards contributor at RSNA, with a broader commitment to reproducible AI, responsible clinical translation, and research-capacity building for Peruvian and international students.