At the Society for Imaging Informatics in Medicine, 2019, NVIDIA released the Clara Train SDK for General Availability. The SDK is available for download now from NGC.
This version includes:
- APIs to add AI-assisted annotation to any medical viewer with new capabilities like Auto-Annotation and interactive annotation modes. Also includes Annotation Server that makes pre-trained models available to the client application; client APIs make integration with your medical viewer application seamless. Annotation Server and Pre-trained models are available on NGC and the APIs are hosted on Github.
- These capabilities are already integrated into the latest MITK workbench plugin.
- The SDK provides capabilities to use techniques like Transfer learning to adapt or train deep learning models from scratch, enabling Data Scientists to bring their own model architectures and run workflows, this is made possible through a unified foundation of Medical Model Archive (MMAR)
- The MMAR provides a model development environment; defines a standard structure for storing and organizing all artifacts produced during the model development life cycle.
- MMAR includes NVIDIA pretrained models based on AH-Net, DenseNet, ResNet, Dextr3D packaged as complete 2D/3D model applications for organ based segmentation, classification and annotation.
With the NVIDIA Clara AI toolkit, NVIDIA is committed to lower the barrier to adoption of AI in medical Imaging and has partnered with the American College of Radiology to enable thousands of radiologists to create and use AI in their own facilities, with their own data, across a vast network of thousands of hospitals.
Additional background: American College of Radiology