As Nvidia continues its focus on the medical imaging space, last night the hardware company unveiled Clara AI, a toolkit for radiologists, which includes 13 classification and segmentation artificial intelligences and software tools. 

At yesterday’s Nvidia’s GTC conference keynote address, the company’s CEO Jensen Huang discussed how artificial intelligence could be the key to helping radiologists save precious time and resources.

“So the question is how do we apply deep learning to enable all of these radiologists so that when they are doing a report there is an assistant sitting next to them helping them along,” Huang asked. “Meanwhile recognizing that there is no way that one institution will to train all of the data networks for all of these rare diseases, many of them happen very infrequently and many of them have experts and specialists in the field.”

Huang doesn’t propose that Nvidia come up with every tool needed in the space. Instead the company would help radiologists develop the tools needed. 

“So what we deiced to do was instead of being the one company to solve it all, we would help them create tools and put them in the hands of radiologist around the world,” he said. “We would give them AI tools and then give them an AI infrastructure.”

One of the focus areas of the new technology is assisting in annotating and labeling large data sets. The company said that it could alleviate the time-consuming task of labeling data sets. Instead Clara AI could help structure data sets. 

“We worked with radiologist and research hospitals, and we dedicated ourselves to making amazing pre-trained models,” he said. “The second thing was we got the tool that was basically an AI that is going to create AIs. It helps with assistance of essentially annotating… and then we can do transfer models. We fine tune these models with data that you compose. So you download a pre-trained model. You annotate your data, you then fine tune that model and then recompile that into a model that you can use.”

The idea is that hospitals and research centers can take this tool and apply it to fit their own data and needs. So far Ohio State University, University of California San Francisco and the National Institutes of Health have employed Nvidia’s Clara AI. 

This Clara AI is announcement is an expansion of the AI platform Clara, which uses AI to create a virtual medical imaging platform, and was announced at last year’s conference. 

Why it matters 

Physicians are in high demand but the forecast looks bleak for filling the need. According to the Association of American Medical Colleges, the US alone is facing a shortage of over 120,000 doctors by the year 2030. Increasingly technologists have been looking to use AI and machine learning to help lighten doctor’s work loads. 

“There are 100,000 radiologists in the United States alone we also have the best trained radiologist in the world. They used to be able to look at [an image] and study it for 40 minutes. Now they have four, the pressure on them is incredible,” Huang said. “Early detection is the best way to stop something before it gets worse. So the question is: How do we apply deep learning to enable all of these radiologists so that when they are doing a report there is an assistant sitting next to them helping them along?”

What's the trend 

This is by no means Nvidia’s first launch in digital health. At last year’s GTC conference, the company announced the first version of Clara. 

The company has also been making deals with major research institutions. For example, in October Nvidia made an announcement that it is partnering with King’s College London to deploy the Clara platform as well as Nvidia’s DGX-2. 

Additionally, the company has been involved in medical imaging and partnering with big names in the healthcare industry like Massachusetts General Hospital and the Mayo Clinic.