Medical transforms
Spatial and intensity pipelines are designed for volumes, not photos.
Open source · Medical imaging
MONAI
Medical images are often 3D, anisotropic, and sparse. MONAI provides transforms, networks, and metrics designed for that reality.
- GitHub stars
- 8.7K
- Language
- Python
- License
- Apache-2.0
- Runs on
- PyTorch · CUDA
01
/ 03What it is
Deep learning adapted to clinical imaging data.
Medical images are often 3D, anisotropic, and sparse. MONAI provides transforms, networks, and metrics designed for that reality.
Domain networks
UNet and SwinUNETR reference architectures are ready to train.
Clinical metrics
Dice, Hausdorff, and surface distance are implemented correctly.
02
/ 03Capabilities
Building blocks for clinical models.
Compose data preparation, training, evaluation, and deployment.
Volume transforms
Composable preprocessing for 3D images, labels, and metadata.
Training workflows
Supervised and self-supervised recipes with reproducible splits.
Deployment tools
Package models for inference and clinical viewing tools.
03
/ 03How to use
Install and compose a pipeline.
Install MONAI beside the PyTorch build that matches your hardware.
01
Install MONAI
Use the PyTorch build that matches your CUDA environment.
terminal
pip install monai02
Compose transforms
Chain transforms and apply them to volume data.
transforms.py
from monai.transforms import Compose, EnsureChannelFirst, ScaleIntensity
transform = Compose([
EnsureChannelFirst(),
ScaleIntensity(),
])Keep exploring
More open-source projects
BioinformaticsBiopython
Python tools for computational biology — read sequence files, query NCBI, parse protein structures, and build phylogenetic trees.
PythonBiopython License5.2K
View Project
Scientific computingNumPy
The foundation of scientific computing in Python — N-dimensional arrays, broadcasting, and fast vectorized math.
Python / CBSD-3-Clause32.7K
View Project
Local AIOllama
Run open-source language models on your own laptop. Pull a model, chat in the terminal, or call it at localhost:11434.
GoMIT181K
View Project