At RSNA, Radiology Poised for Transformation with AI

by Chris Scotto DiVetta

If you want to know what’s next for the medical imaging industry, you head to RSNA — the annual meeting of the Radiological Society of North America.

Later this month, 50,000 radiology experts from across North America will convene at RSNA 2018, in Chicago, to discuss their biggest challenges — and AI will once again be in the spotlight.

From finding operational efficiencies to increasing diagnostic accuracy, AI is already having a big impact on radiology. And it’s being enabled by the arrival of powerful computing solutions designed for medical imaging.

The NVIDIA Clara platform, unveiled in September, combines hardware and software for the development and scalable deployment of applications and radiology workflows. It includes the Clara SDK, a set of GPU-accelerated libraries and containers for computing, visualization and AI that medical-application developers can use to create applications that deploy on device, on premise or in the cloud.

The revolutionary NVIDIA Clara AGX computing architecture is based on the NVIDIA Xavier AI computing module and NVIDIA Turing GPUs.

Mission-critical tasks must be done at the point of care. NVIDIA created the Clara AGX computing architecture to help the next generation of medical instruments address the challenge of processing the massive sea of data — tens to thousands of gigabytes worth — generated each second, so it can be interpreted by doctors and scientists.

Achieving this level of supercomputing has traditionally required three computing architectures: FPGAs, CPUs and GPUs. Clara AGX simplifies this to a single, GPU-based architecture that delivers the world’s fastest AI inferencing on NVIDIA Tensor Core GPUs.

To tackle machine learning challenges, NVIDIA launched RAPIDS, open-source software that drastically increases the productivity of data scientists by speeding up their workflows. RAPIDS will accelerate the process of realizing operational efficiencies from hospital data lakes.

From Training to Deployment

The medical imaging industry is constantly changing, which is why RSNA is big on training and education. Radiologists, researchers and healthcare IT experts attended more than 1,000 sessions at last year’s Deep Learning Classroom, hosted by NVIDIA’s Deep Learning Institute. The conference also debuted a dedicated Machine Learning Pavillion.

RSNA deep learning classroom
RSNA attendees are expected to take more than 1,400 sessions at the Deep Learning Classroom.

The Deep Learning Classroom continues this year with advanced topics aimed at bringing AI to radiology imaging, including the latest methods for increasing sparse datasets through data augmentation, advanced segmentation techniques and multiparametric classification.

NVIDIA will also be showcasing the ease of machine learning and deep learning for medical imaging, from training to deployment.

  • Attend hands-on workshops for practicing radiologists at the Deep Learning Classroom hosted by the Deep Learning Institute. Bring your laptop to work with machine learning tools and write algorithms.
  • Experience interactive demos showcasing the latest AI technology for healthcare at NVIDIA’s booth in the Machine Learning Pavilion – North Hall 3, booth 6568.
  • See the latest AI applications for medical imaging from NVIDIA’s healthcare startup partners with daily talks and demos in the NVIDIA booth.
  • Learn how to get started with deep learning at our talk on Tuesday, Nov. 27, from 8:30-10 a.m. in room S406A. We’ll also be presenting about “Commercial Development & Deployment of Deep Learning Technology” from 4:30-6 p.m. on Tuesday in room E451B.
  • Hear about the era of intelligent instruments from Kimberly Powell, vice president of Healthcare at NVIDIA, during her presentation at the Machine Learning Theater (ML54 Machine Learning Showcase North Hall). Her talk, “Towards Intelligent Healthcare,” takes place on Thursday, Nov. 29, from 12:30-12:50 pm.

For more information about our activities, visit the NVIDIA at RSNA website.