MONAI
Project MONAIPyTorch framework for deep learning on medical images, part of the PyTorch Ecosystem. It supplies medical-imaging transforms, losses and metrics, a large set of network architectures (UNet, UNETR, SwinUNETR, SegResNet, VISTA3D and more), training workflows and model bundles, and Auto3DSeg, which analyzes a dataset, trains several segmentation algorithms and ensembles them with minimal user input. It began at NVIDIA and King's College London and is steered by an advisory board drawn from its member institutions.
Openness
5 high confidence- license
- Apache-2.0(OSI)
- source
- public(github.com/Project-MONAI/MONAI)
- core features withheld
- no — community project steered by an advisory board
Apache-2.0 and built from the public repository. MONAI is run in the open by a community of more than thirty institutions under an advisory board, and no edition of it is sold.
- https://project-monai.github.io/about.html recorded 2026-09-27
About page: "Started by NVIDIA and King's College London, it has grown into a community of 30+ institutions"; "The Advisory Board sets MONAI's strategic direction"; "MONAI is built in the open: code, governance, and roadmap."
- https://raw.githubusercontent.com/Project-MONAI/MONAI/HEAD/LICENSE recorded 2026-09-27
LICENSE is the Apache License, Version 2.0.
- https://raw.githubusercontent.com/Project-MONAI/MONAI/HEAD/README.md recorded 2026-09-27
README: "MONAI is a PyTorch-based, open-source framework for deep learning in healthcare imaging, part of the PyTorch Ecosystem."
- https://ungh.cc/repos/Project-MONAI/MONAI/files/dev recorded 2026-09-27
Repository tree of 1,585 paths with no ee, enterprise, commercial or pro directory.
Adoption
3 high confidenceMeasured on monthly PyPI downloads of the monai package.
- https://pypistats.org/api/packages/monai/recent recorded 2026-09-27
last_month 824,386 downloads of monai
Capability
5 medium confidenceMost of MONAI is a toolkit for building medical-imaging models by hand, but its Auto3DSeg runs the whole job for 3D segmentation - data analysis, algorithm choice, training, tuning and ensembling - from a single line of code, the automation AutoGluon provides for tables. That automation covers segmentation only.
- https://monai.readthedocs.io/en/latest/apps.html recorded 2026-09-27
Auto3DSeg AutoRunner: "An interface for handling Auto3Dseg with minimal inputs ... The users can run the Auto3Dseg with default settings in one line of code."; output includes "data statistics analysis report", "algorithm definition files", "training results (checkpoints, accuracies)" and "the predictions on the testing datasets from the final algorithm ensemble"; "automatic hyperparameter optimization".
- https://monai.readthedocs.io/en/latest/networks.html recorded 2026-09-27
Network reference lists blocks, layers and nets including UNet, UNETR, SwinUNETR, SegResNet, DynUNet, VISTA3D, MedNeXt and ViT.
Verified 2026-09-27