MONAI (Medical Open Network for AI) is an open-source, PyTorch-based toolkit for deep learning in healthcare imaging. Backed by a large academic and industry consortium, it provides the domain-specific building blocks medical imaging projects need, from data loading for DICOM and NIfTI files to 3D segmentation networks and clinical evaluation metrics. The project has about 8,800 stars on GitHub and is actively developed in 2026.
Medical images are not ordinary photos: they are 3D or 4D volumes with specialized formats, spacing and orientation metadata. MONAI handles those realities with imaging-specific transforms, readers and writers, alongside reference implementations of state-of-the-art networks for tasks like organ and tumor segmentation, classification and registration. It also covers the surrounding workflow โ data engineering, experiment management, evaluation with metrics such as Dice scores, and deployment tooling through the wider MONAI ecosystem โ so research code can move toward clinical applications. The wider project includes MONAI Label, an intelligent image-labeling tool that speeds up annotation with AI assistance, Auto3DSeg for automatically configuring strong segmentation pipelines, and a model bundle format for packaging trained models with their metadata for sharing and deployment. MONAI originated from work by NVIDIA and King’s College London and is now governed as a community open-source project, with contributions from hospitals, universities and companies across the medical-imaging world.
MONAI is completely free and open source under the Apache-2.0 license.
MONAI is for medical imaging researchers, data scientists and healthcare AI teams building deep-learning models for radiology, pathology and other clinical imaging tasks.