---

name: "nnunet-segmentation" description: "Medical image segmentation with nnU-Net's self-configuring framework — auto-selects architecture, preprocessing, training for any modality. CT, MRI, microscopy, ultrasound in 2D, 3D full-res, 3D low-res, cascade. Pipeline: convert → plan/preprocess → train (5-fold CV) → best config → predict → ensemble. Use when classical segmentation fails and annotated data exists." license: "Apache-2.0"

nnU-Net Automated Medical Image Segmentation

Overview

nnU-Net (no-new-Net) is a self-configuring deep learning framework for biomedical image segmentation. Given a labeled training dataset, nnU-Net automatically determines the optimal network architecture (2D, 3D full-resolution, or 3D cascade), preprocessing steps (resampling, normalization, patch size), training schedule, and post-processing. It consistently achieves state-of-the-art performance across diverse imaging modalities and anatomical structures without manual hyperparameter tuning. nnU-Net v2 (nnunetv2) is the current release with a Python API for inference alongside the standard CLI.

When to Use

• Segmenting anatomical structures in CT or MRI scans (organs, tumors, lesions) when you have 20+ annotated training cases
• Automating cell or nucleus segmentation in 3D fluorescence or electron microscopy volumes
• Establishing a strong baseline for any new segmentation challenge without manually tuning a U-Net
• Running inference on new images using a pretrained nnU-Net model from a published challenge
• Comparing segmentation methods: nnU-Net's auto-configured ensembles serve as a rigorous baseline
• Building production segmentation pipelines where training is done once and inference is repeated on many new cases
• Use Cellpose (cellpose-cell-segmentation) instead for 2D fluorescence cell segmentation without labeled training data; nnU-Net requires annotated training cases
• Use SimpleITK (simpleitk-image-registration) instead for rule-based segmentation with classical thresholding and region growing on images where deep learning training data is unavailable

Prerequisites

• Python packages: nnunetv2>=2.2, torch>=2.0 (with CUDA for GPU training)
• Data requirements: Images in NIfTI format (.nii.gz); binary or multi-class label masks; minimum 20 training cases recommended (50+ for best performance)
• Environment variables: nnUNet_raw, nnUNet_preprocessed, nnUNet_results must be set
• Hardware: GPU with 8+ GB VRAM recommended for training; CPU inference is supported but slow
• Environment: Python 3.9+; Linux or macOS (Windows supported but not officially recommended)

pip install nnunetv2

# Verify installation
nnUNetv2_train --help

# Set required environment variables (add to ~/.bashrc or ~/.zshrc)
export nnUNet_raw=/data/nnUNet_raw
export nnUNet_preprocessed=/data/nnUNet_preprocessed
export nnUNet_results=/data/nnUNet_results

mkdir -p $nnUNet_raw $nnUNet_preprocessed $nnUNet_results

Quick Start

# Minimal end-to-end pipeline: convert dataset → preprocess → train → predict
# Assumes dataset is in Medical Segmentation Decathlon format

# 1. Set environment
export nnUNet_raw=/data/nnUNet_raw
export nnUNet_preprocessed=/data/nnUNet_preprocessed
export nnUNet_results=/data/nnUNet_results

# 2. Convert Medical Segmentation Decathlon dataset (dataset ID 7 = Pancreas)
nnUNetv2_convert_MSD_dataset -i /data/Task07_Pancreas -overwrite_id 7

# 3. Plan preprocessing and verify dataset integrity
nnUNetv2_plan_and_preprocess -d 7 --verify_dataset_integrity

# 4. Train 3D full-res model, fold 0 (of 5-fold cross-validation)
nnUNetv2_train 7 3d_fullres 0 --npz

# 5. Predict on new images
nnUNetv2_predict -i /data/test_images/ -o /data/predictions/ -d 7 -c 3d_fullres -f 0
echo "Segmentation predictions saved to /data/predictions/"

Workflow

Step 1: Prepare Dataset in nnU-Net Format

nnU-Net requires images in NIfTI format organized in a specific directory structure with a dataset.json descriptor.

# Dataset directory structure (Dataset007_Pancreas as example):
# $nnUNet_raw/
# └── Dataset007_Pancreas/
#     ├── dataset.json          ← metadata descriptor
#     ├── imagesTr/             ← training images
#     │   ├── pancreas_001_0000.nii.gz   (0000 = channel/modality index)
#     │   └── pancreas_002_0000.nii.gz
#     ├── labelsTr/             ← training segmentation masks
#     │   ├── pancreas_001.nii.gz
#     │   └── pancreas_002.nii.gz
#     └── imagesTs/             ← test images (no labels required)
#         └── pancreas_101_0000.nii.gz

# For Medical Segmentation Decathlon datasets, convert automatically:
nnUNetv2_convert_MSD_dataset -i /data/Task07_Pancreas -overwrite_id 7

echo "Dataset 7 created at $nnUNet_raw/Dataset007_Pancreas/"

import json
from pathlib import Path
import shutil

# Create dataset.json for a custom dataset (single CT modality, 2-class segmentation)
dataset_id = 8
dataset_name = f"Dataset{dataset_id:03d}_MyOrgan"
dataset_dir = Path(f"{dataset_name}")
(dataset_dir / "imagesTr").mkdir(parents=True, exist_ok=True)
(dataset_dir / "labelsTr").mkdir(parents=True, exist_ok=True)
(dataset_dir / "imagesTs").mkdir(parents=True, exist_ok=True)

dataset_json = {
    "channel_names": {
        "0": "CT"           # for MRI: "0": "T1", "1": "T2" (multi-modal = multiple channels)
    },
    "labels": {
        "background": 0,
        "organ": 1          # add more classes: "tumor": 2, "vessel": 3
    },
    "numTraining": 50,      # number of training cases
    "file_ending": ".nii.gz"
}

with open(dataset_dir / "dataset.json", "w") as f:
    json.dump(dataset_json, f, indent=2)

print(f"Created dataset.json with {dataset_json['numTraining']} training cases")
print(f"Labels: {dataset_json['labels']}")
print(f"Channels: {dataset_json['channel_names']}")
print(f"\nPlace training images as: imagesTr/case_NNN_0000.nii.gz")
print(f"Place training labels as:  labelsTr/case_NNN.nii.gz")

Step 2: Plan and Preprocess

nnU-Net analyzes the dataset fingerprint (image intensities, spacings, sizes) to automatically configure architectures and preprocessing. This step can take 30–120 minutes for large datasets.

# Plan preprocessing for dataset 7 with integrity checks
nnUNetv2_plan_and_preprocess -d 7 --verify_dataset_integrity

# This creates:
# $nnUNet_preprocessed/Dataset007_Pancreas/
# ├── nnUNetPlans.json           ← architecture configurations (patch size, batch size, etc.)
# ├── dataset_fingerprint.json   ← image statistics used for auto-configuration
# ├── nnUNetPlans_2d/            ← preprocessed 2D slices
# ├── nnUNetPlans_3d_fullres/    ← preprocessed 3D full-resolution patches
# └── nnUNetPlans_3d_lowres/     ← preprocessed 3D low-resolution patches (if applicable)

echo "Preprocessing complete. Plans stored in $nnUNet_preprocessed/Dataset007_Pancreas/"

import json
from pathlib import Path
import os

# Inspect the auto-generated plans to understand what nnU-Net configured
preprocessed_dir = Path(os.environ["nnUNet_preprocessed"])
plans_file = preprocessed_dir / "Dataset007_Pancreas" / "nnUNetPlans.json"

with open(plans_file) as f:
    plans = json.load(f)

print("nnU-Net auto-configuration summary:")
print(f"  Dataset name: {plans['dataset_name']}")
print(f"  Original spacing (mm): {plans['original_median_spacing_after_transp']}")

for config_name, config in plans["configurations"].items():
    print(f"\n  Configuration: {config_name}")
    print(f"    Patch size:     {config.get('patch_size', 'N/A')}")
    print(f"    Batch size:     {config.get('batch_size', 'N/A')}")
    print(f"    Spacing:        {config.get('spacing', 'N/A')}")
    print(f"    Network arch:   {config.get('network_arch_class_name', 'N/A')}")

Step 3: Train (5-Fold Cross-Validation)

nnU-Net uses 5-fold cross-validation by default. Train all 5 folds per configuration for the best ensemble performance, or train fold 0 only for a quick first model.

# Train 3D full-resolution configuration, fold 0 (on GPU)
# --npz saves softmax predictions for ensembling
nnUNetv2_train 7 3d_fullres 0 --npz

# Train all 5 folds (submit as 5 parallel jobs on a cluster)
for fold in 0 1 2 3 4; do
    nnUNetv2_train 7 3d_fullres $fold --npz
done

# Also train 2D configuration for potential ensemble
nnUNetv2_train 7 2d 0 --npz

echo "Training outputs in $nnUNet_results/Dataset007_Pancreas/"

# Resume training from a checkpoint (if interrupted)
nnUNetv2_train 7 3d_fullres 0 --npz --c

# Training on CPU only (slow — for testing only)
nnUNetv2_train 7 3d_fullres 0 --npz --device cpu

# Monitor GPU memory and training progress
watch -n 10 nvidia-smi

# Check training log for loss curves
tail -f $nnUNet_results/Dataset007_Pancreas/nnUNetTrainer__nnUNetPlans__3d_fullres/fold_0/training_log.txt

Step 4: Find Best Configuration

After training, nnU-Net evaluates all trained configurations and their ensembles to recommend the best single model or ensemble.

# Determine which configuration or ensemble performs best on validation folds
# Must have trained multiple folds (0-4) for accurate comparison
nnUNetv2_find_best_configuration 7 -c 2d 3d_fullres

# Output example:
# Best configuration: 3d_fullres
# Best ensemble: 2d + 3d_fullres (Dice = 0.823 vs 3d_fullres alone = 0.814)
# Recommended: ensemble of 2d + 3d_fullres

# Also run post-processing determination
# (removes small connected components if it helps validation Dice)
nnUNetv2_apply_postprocessing --help

import json
from pathlib import Path
import os

# Parse cross-validation results to see per-fold performance
results_dir = Path(os.environ["nnUNet_results"])
summary_file = results_dir / "Dataset007_Pancreas" / \
    "nnUNetTrainer__nnUNetPlans__3d_fullres" / "fold_0" / "validation" / "summary.json"

if summary_file.exists():
    with open(summary_file) as f:
        summary = json.load(f)

    metrics = summary["foreground_mean"]
    print("3D full-res fold_0 validation metrics:")
    for metric_name, value in metrics.items():
        print(f"  {metric_name}: {value:.4f}")
    # Expected output:
    # Dice: 0.81
    # IoU: 0.68
    # HD95: 12.3 (mm)
else:
    print(f"Summary not found at {summary_file}")
    print("Run nnUNetv2_train first to generate validation metrics")

Step 5: Predict on New Images

Run inference on a folder of new images using the trained model.

# Predict using the best single configuration (fold 0 only, fast)
nnUNetv2_predict \
    -i /data/test_images/ \
    -o /data/predictions_3d/ \
    -d 7 \
    -c 3d_fullres \
    -f 0 \
    --save_probabilities    # save softmax probabilities for later ensembling

# Predict using all 5 folds (better performance, slower)
nnUNetv2_predict \
    -i /data/test_images/ \
    -o /data/predictions_allfolds/ \
    -d 7 \
    -c 3d_fullres \
    -f 0 1 2 3 4

echo "Predictions saved to /data/predictions_allfolds/"

import torch
from nnunetv2.inference.predict_from_raw_data import nnUNetPredictor
from pathlib import Path
import os

# Python API for inference — useful for integration into custom pipelines
results_dir = os.environ["nnUNet_results"]
model_folder = f"{results_dir}/Dataset007_Pancreas/nnUNetTrainer__nnUNetPlans__3d_fullres"

# Initialize predictor
predictor = nnUNetPredictor(
    tile_step_size=0.5,      # overlap fraction between tiles (higher = better quality, slower)
    use_gaussian=True,       # Gaussian weighting at tile edges (reduces boundary artifacts)
    use_mirroring=True,      # test-time augmentation with flips (improves accuracy)
    device=torch.device("cuda", 0),  # use GPU 0; set to torch.device("cpu") for CPU
    verbose=False,
    allow_tqdm=True,
)

predictor.initialize_from_trained_model_folder(
    model_folder,
    use_folds=(0,),          # which folds to use; (0,1,2,3,4) for ensemble of all folds
    checkpoint_name="checkpoint_best.pth",  # or "checkpoint