Imaging providers worldwide may soon have access to a single source of AI-powered continuous updates for their existing install bases in the form of a single, virtual supercomputer.

Nvidia Corporation, a technology enterprise known for its work in gaming and self-driving cars, revealed its intent to build such an entity last month at its GPU Technology Conference, demonstrating with ultrasound and CT that a virtual supercomputer could utilize AI to capture patient data from imaging modalities as well all other resources to create and deploy imaging applications anywhere. Nvidia refers to its initiative as Project Clara.

“It’s not dissimilar to thinking of it as your iPhone,” Kimberly Powell, vice president of healthcare at Nvidia, told HCB News. “Over time, a new application comes to your iPhone and all of a sudden, it gives you this new capability, but your iPhone didn’t have to get upgraded. It’s just a new application.”
Historically, practices have been bound by the computations available in their scanners. Of the three million imaging devices installed throughout the world, only 100,000 are sold each year, making the feat of updating them all a 30-year endeavor if attempted.

With Clara, physicians will immediately have access to the latest computational applications for a variety of tasks, from imaging reconstruction to imaging processing to sensor compression, enabling greater quality for interpretations without experiencing interruptions in workflow.

The idea behind the platform envisions a combination of GPU and cloud-based technology with current computer architecture to form a finished product capable of evaluating data, whether it's graphics-oriented, computer-oriented, or AI-oriented or a combination of all three.

To develop Clara, Nvidia has teamed up with large incumbent imaging vendors as well as up-and-coming startup companies and world renowned research hospitals, such as Siemens Healthineers, Massachusetts General Hospital, Samsung, NYU School of Medicine, and Fujifilm.

It also announced two partnerships at this year’s RSNA in November, one with GE and the other with Nuance.

Such capabilities hold the potential to extend the life of medical imaging equipment, grow levels of scale and capabilities in medical imaging and convince manufacturers to rethink their designs and creation processes for medical devices and software.

“Our goal is to have something at RSNA that really illustrates how you can virtualize a medical instrument and show how this platform can be used to not only make our install base more intelligent, but enable new, innovative devices to come to market as well,” said Powell.

The project is currently in the “definition phase”, with researchers and developers deciphering how to compose it from today’s computer architecture and technology.

A time frame for its release has not yet been determined.

CT Homepage