Nvidia AI creates new programs for radiology, drug discovery

Nvidia has collaborated with two different medical groups to create new artificial intelligence-driven programs to help with diagnoses and drug discovery.

Nvidia and the American College of Radiology (ACR) worked to create a software platform that will allow more than 38,000 radiology professionals to use AI and machine learning in their work diagnosing disease and injury using medical imaging. The software comes after a three-month pilot program by Nvidia and ACR.

The new software integrates Nvidia Clara AI toolkit into the ACR Data Science Institute's ACR AI-LAB. It allows radiologists to build, share, adapt and validate AI algorithms. The toolkit is part of the Nvidia Clara developer platform, which is a platform to create data and algorithm pipelines.

It was specifically designed to enable software-defined medical instruments and intelligent workflows, and consists of libraries for data and image processing, AI model processing and visualization. The toolkit also includes libraries for data annotation, model training, model adaptation, model federation and large-scale deployment.

Nvidia initially had a pilot program with the Ohio State University and the Center for Clinical Data Science, a collaboration between Massachusetts General Hospital and Brigham and Women's Hospital. Ohio State and the Center for Clinical Data Science helped Nvidia and ACR find which pathways were necessary to enable facilities to work together without sharing sensitive patient data.

The Center for Clinical Data Science was able to develop a pretrained model, which Ohio State was able to import using the Nvidia AI healthcare toolkit.

The initial version of ACR AI-LAB will debut at the 2019 ACR Annual Meeting in Washington, D.C. from May 18 to 22. ACR plans to provide online access and sample data from publicly available patient data sets soon after.

Nvidia also joined forces with the Accelerating Therapeutics for Opportunities in Medicine (ATOM) to create an AI-based drug discovery program. ATOM was formed in October 2017 with the goal of transforming drug discovery from a slow, high-failure process to a "rapid, integrated and patient-centric model," according to ATOM.

To do this, ATOM is using Nvidia's GPU-accelerated supercomputers. The companies will work together to advance the ATOM platform and make it more accessible. The goal is to use Nvidia's AI, computing and resources with ATOM's predictive model to reduce the time and cost to discover new drugs.