Research team at the Clinical Data Science Center of Massachusetts General Hospital

Data scientists on the research team trained to identify regular patterns and malformations in medical images in just eight months after receiving NVIDIA's first-generation DGX-1 AI supercomputer last December. In addition, dozens of medical training algorithms have been developed to date, covering not only radiology but also specialized medical fields such as cardiology, ophthalmology, dermatology, and psychiatry.

The research team plans to advance groundbreaking research aimed at developing multiple training algorithms and directly delivering AI performance to medical professionals using the NVIDIA DGX Station, the world's first personal AI supercomputer scheduled for release this month, and the NVIDIA DGX-1, a supercomputer based on the Volta architecture.

Neural networks trained on the DGX-1 system built within the research team's data center continuously acquire numerous medical images from around the world. This not only provides superhuman performance in identifying images at the pixel level but also analyzes other data scores at an incredible speed, supporting faster and more accurate disease diagnosis by medical professionals in MRI, CAT scans, X-rays, and more.

"Mark Michalski, Executive Director of the Clinical Data Science Center at Massachusetts General Hospital, explained, 'Amidst a flood of data—including laboratory reports, MRIs, CAT scans, and family health records—practitioners today face significant difficulties in making rapid decisions. Therefore, possessing technology that can support their efforts to make such decisions will lead to highly innovative results.'"

NVIDIA, together with a research team from the Clinical Data Science Center at Massachusetts General Hospital, plans to showcase cutting-edge medical imaging research at the upcoming MICCAI 2017 conference (Medical Image Computing and Computer Assisted Interventions Conference) in Quebec, Canada.