optional extras as needed

Overview

MONAI is the standard PyTorch-based framework for medical imaging deep learning. Use it for segmentation, classification, registration, preprocessing, and training pipelines on DICOM, NIfTI, and other medical image formats.

Installation

uv pip install monai
# optional extras as needed
uv pip install monai[all]

Core strengths

MONAI gives you:

• medical-image-aware transforms
• domain-specific architectures like UNet, UNETR, SwinUNETR
• losses like DiceLoss and DiceCELoss
• metrics like Dice and Hausdorff distance
• dataset and engine utilities for training loops

Basic 3D segmentation model

import torch
from monai.networks.nets import UNet
from monai.networks.layers import Norm

model = UNet(
    spatial_dims=3,
    in_channels=1,
    out_channels=3,
    channels=(16, 32, 64, 128, 256),
    strides=(2, 2, 2, 2),
    num_res_units=2,
    norm=Norm.BATCH,
).cuda()

Preprocessing transforms