Nvidia rolls out AI models designed to behave like radiologists interpreting images
Nvidia—a leader in artificial intelligence computing—is rolling out a series of helpful tools targeted at advancing explainable AI.
Nvidia Clara houses a series of “models, tools, and recipes that are built for accelerating scientific discovery, analyzing medical images, and providing a foundational understanding of human health, biology and chemistry,” a recent Nvidia blog explained. Specifically, Clara Reason will introduce a series of multimodal chain-of-thought models that have been designed to behave like a radiologist interpreting images. Clara Reason provides step-by-step explanations on how it arrives at conclusions, increasing trust in its reasoning and assessments.
The Clara NV-Reason-CXR-3B model—a vision language model designed to systematically evaluate chest radiographs through the same process radiologists use—is among the first explainable AI offerings. It provides a step-by-step diagnostic analysis, anatomical review, both normal and abnormal findings and a suggested differential diagnosis. What’s more, the model gives follow-up recommendations, generates a structured report and offers a multistep follow-up chat feature.
Nvidia collaborated with the National Institutes of Health, Children’s Hospital of Philadelphia, and VinBrain to create the chain-of-thought dataset. The extensive dataset provides multiple pages of explanations for each individual image, rather than single reports or image labels. This offers users ample data to consider when evaluating the tool’s conclusions.
Clara Reason’s capabilities have been evaluated by board certified radiologists. Nvidia says it can “act as a co-pilot" for radiologists by enhancing their workflows and offering decision support.
“The CXR reasoning model is an amazing opportunity for assisting not only referring doctors but also patients who would like to learn more about the thought process of establishing differential diagnoses using imaging findings from all anatomic structures covered in the field of view, along with patients’ clinical information and symptoms,” Ismail Baris Turkbey, MD, FSAR, a senior clinician with the National Institutes of Health, said in the blog. “Additionally, this novel tool has significant potential to serve as an educational assistant for trainees in radiology and medicine.”
For detailed information on how to use the open models for research and development purposes, click here and scroll down.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.