MONAI: An Open-Source Imaging Framework To Accelerate AI in Healthcare, Ready For Production Using NVIDIA Clara Application Framework

3342
Source: https://monai.io/start.html

MONAI (the Medical Open Network for AI) is a PyTorch-based framework that contributes to AI development for healthcare issues. It is looking to address medical imaging issues with industry-specific data handling, high-performance training workflows, and reproducible reference implementations of state-of-the-art approaches. MONAI is ready for production as of April 2020, with the NVIDIA Clara application framework’s upcoming release.

Built with the updated Clara offering, MONAI is going to include 20 pre-trained models. It also consists of the ones recently developed for COVID-19 and the latest training optimizations on NVIDIA DGX A100 GPUs to enhance the speed sixfold.

MONAI is said to be the “PyTorch of healthcare,” reducing the gap and encouraging the collaboration between data scientists and clinicians. DKFZ- Deutsches Krebsforschungszentrum (German Cancer Research Center), King’s College London, Mass General, Stanford, and Vanderbilt are recognized institutions that will adopt MONAI for imaging. MONAI is being used in various places, and its adoption by the healthcare ecosystem is tremendous.

New Features

NVIDIA Clara brings the latest advancements in AI-assisted annotationfederated learning, and production deployment. 

  • The latest version allows radiologists to label complex 3D CT data in one-tenth of the clicks with a new model called DeepGrow 3D. Where the traditional method of segmenting an ‘organ’ or a ‘lesion’ image by image or slice by slice used to consume much time, this version allows users to segment with far fewer clicks.
  • NVIDIA Clara’s AI-assisted annotation tools and DeepGrow 3D feature are now integrated with Fovia Ai’s FAST AI Annotation software. This integration can be used in labeling training data and assisting radiologists when reading. 
  • AI-assisted annotation is used for accessing extensive radiology datasets. It has been used to label the public COVID-19 CT dataset recently published by The Cancer Imaging Archive at the US National Institutes of Health. This labeled dataset was then used in the MICCAI-endorsed COVID-19 Lung CT Lesion Segmentation Challenge.
  • Clara Federated Learning recently played a significant role in the research collaboration of 20 hospitals worldwide by developing a generalized AI model for COVID-19 patients. 

Project MONAI, jointly with the rest of the NVIDIA Clara ecosystem, is expected to improve patient care and optimize hospital operations.

Source: https://blogs.nvidia.com/blog/2020/11/30/monai-ai-imaging-framework/

Github: https://github.com/Project-MONAI