The NVIDIA IGX Platform for Medical AI Use Cases
Organizations from various industries are exploring ways to increase the use of automation to increase productivity, efficiency and safety. Computers can speed up this process by learning to recognize patterns and reliably perform the same tasks over and over again. However, reality is not deterministic, and the breadth of human activity encompasses various tasks and contexts that rules and programs cannot adequately capture.
Edge AI refers to the practice of performing AI calculations at the edge of the network, closer to the user and the actual data, rather than in a centralized location such as a service provider’s server farm on the network. cloud or a data center of a private company.
Edge AI has advanced to the point where machines and devices can now function with the “intelligence” of human cognition, wherever they are. AI-powered smart applications can adapt to new situations and learn to successfully execute the same or similar tasks.
With the release of the NVIDIA IGX platform, NVIDIA has brought state-of-the-art security and safety to intelligent machines and human-machine collaboration, making it ideal for application in edge AI medical use cases.
The company has launched IGX for Medical Devices, a computer that will be used in robots, imaging scanners and other patient care devices. “Medical devices can now benefit from the same business model and innovation as autonomous vehicles, to be powered by AI and become software-defined,” said Kimberly Powell, vice president of healthcare at NVIDIA.
NVIDIA machines will complement doctors and nurses. Real-time image display could aid in medical exploration, robotic navigation, or healthcare.
The IGX hardware contains a 250 TOPS Orin module, a 400 Gigabit Ethernet ConnectX-7 connection, and a 600 TOPS RTX Ampere GPU. “More robots and humans will operate in the same scenario,” adds Powell, who explains that commercial operating systems are compatible. Many robotic and automated systems are based on real-time operating systems, executing high-priority functions ahead of less important tasks.
NVIDIA’s perception of real time is 10 to 50 milliseconds off, Powell said. “That is what we intend to give: real-time AI, human perception.” The executive added that NVIDIA will eventually offer a real-time operating system for robotics control.
IGX is paired with Clara HoloScan, a medical software framework that runs robotic surgery and imaging applications. Powell explained that three NVIDIA business partners have adopted IGX with Clara HoloScan. Activ Surgical chose the platform to “process information from its proprietary hyperspectral sensors” to provide surgeons with real-time physical structure visualization and perfusion imaging during surgery.
Moon Surgical chose the platform for its state-of-the-art robotic assisted system, Maestro, which acts as an assistant to the surgeon. Proximie’s telepresence solution in the operating room allows remote participation of the surgeon.
IGX and Clara HoloScan are 60601 Medical Security Certified. NVIDIA provides 62304 Medical Security Certification documentation for embedded medical device software to its partners. “The final app developer goes through FDA approval for the full app and platform, but we get them two-thirds of the way into the platform software,” Powell said.
BioNeMo is part of NeMo LLM, NVIDIA’s latest cloud service for protein folding and drug development. Language, chemistry, and biology are used to predict the qualities of materials or find medicines.
NVIDIA’s Clara platform is now available on the Broad Institute’s Terra platform. The Broad Institute was founded by MIT, Harvard and others. “With sequencing prices dropping from hundreds to $100, we need to efficiently analyze genomic data,” said Kimberly Powell, vice president of healthcare at NVIDIA.
Eddie is an Australian news reporter with over 9 years in the industry and has published on Forbes and tech crunch.