Medical Open Network for AI

The MONAI framework is the open-source foundation being created by 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.

Open Source

MONAI is an open-source project. It is built on top of PyTorch and is released under the Apache 2.0 license.

Standardized

Aiming to capture best practices of AI development for healthcare researchers, with an immediate focus on medical imaging.

User Friendly

Providing user-comprehensible error messages and easy to program API interfaces.

Reproducible

Provides reproducibility of research experiments for comparisons against state-of-the-art implementations.

Easy Integration

Designed to be compatible with existing efforts and ease of 3rd party integration for various components.

High Quality

Delivering high-quality software with enterprise-grade development, tutorials for getting started and robust validation & documentation.

Getting Started

Install monai

You can quickly get started with
the Python Package Index (PyPI)

Or via several
alternative installation approaches

User Guides

MedNIST image classification

MONAI for PyTorch users

Abdominal CT segmentation with 3D UNet
Medical image segmentation tutorial

Fast training with MONAI components
Approximate 12x speedup with CacheDataset, Novograd, and AMP

Workflows

Training and evaluation - classification
Brain MRI classification examples

Training and evaluation - segmentation
Volumetric image segmentation examples

Training and evaluation - MONAI workflows
Image segmentation examples with engine and event-handlers

Training and evaluation - distributed data parallel
Examples based on PyTorch, Horovod, and MONAI workflows

Learn more

View a list of all
MONAI tutorials and examples

New Features in v0.3.0

More components

More domain-optimized networks, loss functions, metrics, and optimizers.

Distributed training

Components fully support distributed data parallel with rich examples.

Auto mixed precision

Enabled PyTorch AMP in workflows, provided rich tutorials and examples.

IO factory

New image readers and LoadImage transform for more medical image formats.

C++/CUDA support

C++/CUDA optimized components and enabled CI/CD for to build.

New examples

Various tutorial notebooks and examples for new features.

User Value Proposition

Entry Level Researcher

Provides ready-to-use deep learning models, walk-through tutorials and examples to get kick-started for AI development on a common foundation.

Advanced Researcher

For advanced academic and translational researchers, MONAI provides modular domain optimized components that can be easily integrated into existing workflows and enables reproducibility of research experiments for comparison.

Repository and Roadmap

Project MONAI

GitHub Repository
Go to repository

Tutorials GitHub Repository
Go to examples and notebooks

Contributing Guidelines
Go to guide

Licensing Guide
Go to Apache 2.0 license

Bug reports and feature requests
Go to issues

Initial Roadmap Schedule

Release Roadmap APR 2020 JUL 2020 OCT 2020 DEC 2020 0.1.0 0.2.0 0.3.0 0.4.0

Contributors

Project MONAI is an initiative originally started by NVIDIA & King’s College London to establish an inclusive community of AI researchers for the development and exchange of best practices for AI in healthcare imaging across academia and enterprise researchers.

Join Project MONAI