May 25, 2022

This week, Nvidia, known for its powerful graphics processing units (GPUs), unveiled an AI-powered medical device development platform. Called the Clara Holoscan MGX, the device delivers processing power that enables medical sensors to process multiple data streams in parallel, train AI algorithms, and visualize biology in real time.

Debuting at Nvidia’s 2022 GTC conference, the Clara Holoscan MGX is an “open, scalable robotics platform,” as CEO Jensen Huang put it in a keynote speech. It is a hardware/software stack designed to connect medical robotic equipment or sensors to AI applications.

How can it work? Take the process of endoscopy as an example. Usually, the doctor inserts a small camera into your body and examines you. Clara Holoscan MGX can directly connect to this camera and process the collected data in real time. This data can then be fed into an AI model that can detect abnormalities, navigate your anatomy, and help surgeons create a treatment plan. (To be clear, these AI models will not be manufactured by Nvidia, they will simply run on their hardware.)

Nvidia is already known for its GPUs, which are especially good for fast, parallel computing. GPUs were once the most familiar to gamers, but they have become the main accelerator for any industry interested in training deep neural networks. Deep neural networks must quickly crack billions of data points by learning to read X-rays. And the resulting models also require a lot of calculations to be used in real time, which is what this platform is designed to provide.

Nvidia is currently a big player in AI because it provides the raw processing power needed for many of these projects and has made it easy with an array of industry-specific hardware and software combinations. For example, NVIDIA is active in the field of self-driving vehicles with projects such as Nvidia Drive, a platform for learning and building autonomous vehicles.

Nvidia has already started making changes to healthcare. The Clara platform, first announced in 2018, was originally designed for seamless medical imaging. The platform has expanded over the years, but the Clara Holoscan MGX platform is mainly aimed at being a one-stop shop.

Kimberly Powell, vice president of healthcare at Nvidia, told gaming-updates that Clara Holoscan is “a complete end-to-end platform. Nvidia Clara Holoscan is to medical devices what Nvidia Drive is to autonomous vehicles,” she said.

Powell says Clara Holoscan’s key innovations are twofold. First, it is designed in accordance with the IEC 62304 standard, the reference process for the secure development of medical software. Then it’s actually loaded with what Huang calls an “insane” amount of computing power.

Together, this combination should allow companies that make or train AI-based medical devices to grow much faster.

Diagram showing the capabilities of the Clara Holoscan MGX.

image credit: NVIDIA

“The Clara Holoscan architecture greatly reduces the technical investment required to market a new medical device or software as a medical device,” Powell said.

There are already enough companies that want to do exactly what Nvidia offers: to combine devices and AI. ActiveSurgical, for example, a startup working on an artificial intelligence surgical endoscope (called ActiveSight), is already taking advantage of Nvidia GPUs and is working on other AI applications based on endoscope data. In doing so, the company was accepted into the Nvidia Inception program, giving it early access to the Clara AGX developer kit. According to the press release, the kit echoes Powell’s claim that Nvidia’s technology can speed up product development.

“The developer kit will reduce the overall development time to market for future ActiveSurgical products, including ActiveSite, over the next two years,” Active Surgical said in a statement.

All features of Clara Holoscan are currently unavailable. During his keynote speech, Huang said that medical technology won’t be available for early access until the first quarter of 2023. At that time, Powell said, hardware pricing will be determined by Nvidia’s ODM partners, while software pricing “will be available.” (This information will most likely appear here.)

For now, the launch of the Clara Holoscan MGX marks Nvidia’s already solid foothold in AI-powered healthcare. In essence, it forms the computational foundation that sits underneath.

And it’s a good place. According to the Stanford AI Index 2022 report, the two largest areas of private investment in artificial intelligence in 2021 are right under the gun: $12.2 billion in data management, processing and cloud computing and $11.29 billion in healthcare. resources and health resources.

Leave a Reply

Your email address will not be published.