Combining Nvidia's Xavier module with Turing GPUs and a new software development kit, Nvidia Clara is intended as a platform that can take the massive amounts of data generated by medical instruments, such as ultrasound and X-ray machines, visualising it, and potentially recommending actions.
"This is a supercomputing problem," said Nvidia vice-president for healthcare, Kimberley Powell. It traditionally requires a combination of CPUs, GPUs and FPGAs, she sad, but Clara does the job with GPUs alone and is scalable from entry-level devices to computationally demanding 3D instruments.
One advantage of Nvidia's approach is that it has the potential to provide a cost-effective way of extending the life of existing instruments by adding software that implements modern techniques.
Early adopters include ImFusion, which is using Clara to create 3D ultrasound images from 2D instruments, and Subtle Medical, which is developing a system that will acquire MRI images with reduced exposure to radiation and lower doses of contrast.
Clara is now at the stage where it is available to early access partners, and a beta release is planned for the second quarter of 2019.