When Seattle-based startup Proprio Vision says its technology for fusing human and computer vision is cutting edge, they mean it.
While steady hands are an asset during surgery, seeing clearly is critical. This is why surgeons depend on a variety of technologies like glasses equipped with magnifying loupes, surgical microscopes, navigation displays and X-ray machines to guide them.
But these standalone devices can’t communicate with or connect to each other. Each provides surgeons a narrow view. So they juggle between devices to prepare for and navigate operations.
Proprio Vision fuses human and computer vision to give surgeons a comprehensive view in immersive, high-definition 3D.
The company compresses the capabilities of multiple bulky and expensive devices into a single light field capturing and rendering system — without giving up imaging features from any component device.
Powered by NVIDIA Quadro GPUs, Propio’s system captures the live surgical field and fuses it with preoperative imaging to create interactive 3D visualizations for surgeons to use in real time.
“Doctors no longer have to look at multiple screens or bring in different machines to get the information they need — our NVIDIA GPU-powered system makes real-time, accurate visualizations possible,” said Proprio CEO Gabriel Jones.
3D Imaging for Surgical Precision
Proprio uses the latest Quadro RTX 6000 GPU and others for tasks like multi-modal image rendering and registration, image processing and correction, 3D geometry reconstruction and CT segmentation.
With the speed and performance of Quadro, Proprio has developed a light field camera array and graphics processing pipeline that captures a high-definition volumetric representation of the surgical field that can be magnified and re-focused anywhere.
Proprio’s custom algorithms render and stream images in vivid 3D, fuse it with preoperative CT and MRI scans, and update the combined scene in real time. Machine learning is integrated for CT segmentation, light field array calibration, depth estimation and tracked user motion analysis.
With this technology, surgeons can use a VR headset to zoom through tissue, around corners and down to microscopic levels, navigating freely and fluidly as they operate.
Propio’s system can also be useful for training future surgeons. Its technology allows doctors to gain familiarity with tools, learn new skills and simulate surgical procedures — all in a virtual environment.
To get the speed and performance they need for real-time visualizations, Proprio moved from CPU to GPU-based rendering. When upgrading from NVIDIA Pascal to Turing architecture, Proprio saw performance increases across various computational tasks, including a 3X faster performance boost for geometry estimation.
“With Proprio’s system combined with the speed and power of Quadro RTX, surgeons can focus on what they do best and perform in the most natural way possible, without any interruptions,” said Jones.
Proprio continues to develop the technology in partnership with neuro and orthopedic surgeons at Seattle Children’s Hospital, the University of Washington School of Medicine and other research hospitals.
Learn More About NVIDIA GPUs in Healthcare
Register now to attend GTC DC.