NVIDIA open sources MONAI (Medical Open Network for AI), a framework developed by NVIDIA and King’s College London for healthcare professionals using best practices from existing tools, including NVIDIA Clara, NiftyNet, DLTK, and DeepNeuro. Using PyTorch resources, MONAI provides domain-optimized foundational capabilities for developing healthcare imaging training in a standardized way to create and evaluate deep learning models.
The MONAI framework is the open-source tool based on Project MONAI. MONAI is a freely available, community-supported, PyTorch-based framework for deep learning in healthcare imaging. It provides domain-optimized foundational capabilities for developing healthcare imaging training workflows in a native PyTorch paradigm.
Modular, open-source solutions provide researchers the flexibility to customize their deep learning development, without needing to replace their existing workflows with an end-to-end system. For example, a researcher can use MONAI code for data preprocessing and transformations and then switch over to an existing AI pipeline for training.
One of the main goals of MONAI is to enable replication of experiments for researchers so that they can share results and build upon each other’s work to advance state of the art.
Features:
- It is flexible for pre-processing of multi-dimensional medical imaging data;
- It provides and supports compositional & portable APIs for ease of integration in existing workflows;
- It provides domain-specific implementations for networks, losses, evaluation metrics and more;
- It has a customizable design for varying user expertise;
- It supports multi-GPU data.
Installation:
pip install monaiTo install from the source code repository:
pip install git+https://github.com/Project-MONAI/MONAI#egg=MONAIGithub: https://github.com/Project-MONAI/MONAI
Web: https://monai.io/
API documentation: https://monai.readthedocs.io/en/latest/
