emorycn2l/emory_robust_wmh repository overview

⁠Emory Robust WMH Segmentation Docker

This Docker image provides a robust and automated pipeline for white matter hyperintensity (WMH) segmentation from heterogeneous clinical MRI, using T1-weighted (T1w) and FLAIR images. The pipeline is designed to handle variability in acquisition protocols and image quality, leveraging deep learning–based inference.

⸻

⁠Authors

Junjie Wu, PhD Deqiang Qiu, PhD

⸻

⁠Overview

The Emory Robust WMH pipeline performs end-to-end processing including:       •     Optional N4 bias field correction       •     Optional T1w-to-FLAIR registration       •     Deep learning–based WMH segmentation using nnU-Net v2

The final output is a WMH segmentation mask written to a user-specified path.

⸻

⁠Usage

Run the container with appropriate volume mounting.

⁠CPU Execution

docker run --rm --shm-size=16g
-v /path/to/data:/data
emorycn2l/emory_robust_wmh
-t /data/T1w.nii.gz
-f /data/FLAIR.nii.gz
-o /data/seg.nii.gz
[--no-n4] [--no-coreg]

⁠GPU Execution

docker run --gpus all --rm --shm-size=16g
-v /path/to/data:/data
emorycn2l/emory_robust_wmh
-t /data/T1w.nii.gz
-f /data/FLAIR.nii.gz
-o /data/seg.nii.gz
--gpu
[--no-n4] [--no-coreg]

⁠Options

      •     -t, --t1 (required): Path to T1-weighted MRI       •     -f, --flair (required): Path to FLAIR MRI       •     -o, --output (required): Output WMH segmentation mask

⁠Optional flags:

      •     --no-n4: Disable N4 bias field correction       •     --no-coreg: Disable T1w-to-FLAIR registration       •     --gpu: Enable GPU inference       •     -h, --help: Display help message

⁠Input

Provide T1w and FLAIR images in NIfTI format. Paths must be accessible within the mounted Docker volume.

⁠Output

WMH segmentation mask (NIfTI format) written to the user-specified output path (e.g., /data/seg.nii.gz)

⁠Methods Summary

The pipeline integrates classical image processing and modern deep learning:       •     Bias field correction: Improves intensity homogeneity       •     Cross-modality registration: Aligns T1w to FLAIR space       •     Segmentation: nnU-Net v2 ensemble models trained on heterogeneous clinical datasets

This design ensures robustness across scanners, protocols, and clinical populations.

⁠License

⁠License for Emory Robust WMH Segmentation

This work, including all original code, configuration, and documentation files provided in this repository or Docker image, is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).

You are free to:

Share — copy and redistribute the material in any medium or format Adapt — remix, transform, and build upon the material Under the following terms:

Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. NonCommercial — You may not use the material for commercial purposes. Full license text: https://creativecommons.org/licenses/by-nc/4.0/legalcode⁠ Summary: https://creativecommons.org/licenses/by-nc/4.0/⁠

⁠Preferred Attribution

If you use this pipeline in your work, please cite:

Wu, J, Brown, JD, Hu, R et al. Benchmark White Matter Hyperintensity Segmentation Methods Fail on Heterogeneous Clinical MRI: A New Dataset and Deep Learning–Based Solutions. Journal of Imaging Informatics in Medicine (2026). https://doi.org/10.1007/s10278-025-01808-9⁠

⁠Third-Party Software and Licenses

This Docker image includes the following third-party components, each governed by its own license:

nnU-Net (v2)       •     Description: Self-configuring deep learning framework for biomedical image segmentation       •     License: Apache License 2.0       •     Source: https://github.com/MIC-DKFZ/nnUNet⁠

⸻

ANTs (Advanced Normalization Tools)       •     Description: Image registration and normalization toolkit (used for N4 bias correction and registration)       •     License: Apache License 2.0       •     Source: https://github.com/ANTsX/ANTs⁠

⸻

SynthStrip (FreeSurfer)       •     Description: Deep learning–based brain extraction tool       •     Part of: FreeSurfer       •     License: FreeSurfer Software License (non-commercial research use)       •     Source: https://github.com/freesurfer/freesurfer⁠

⁠Redistribution Notice

You may redistribute this Docker image or derivative works only for non-commercial purposes and only if you include:

This full LICENSE file A clear description of which components are covered under which licenses All third-party license files and links as required You are responsible for ensuring compliance with all applicable licenses when using or distributing this image.

⁠For Commercial Use

Commercial use of this software or any modified version is prohibited without prior written permission from the authors.

For inquiries, contact: [email protected]⁠