DLMUSE - Deep Learning MUlti-atlas region Segmentation utilizing Ensembles of registration algorithms and parameters
DLMUSE uses a trained nnUNet model to compute the segmentation of the brain into MUSE ROIs from the nifti image of the Intra Cranial Volume (ICV - see DLICV method), oriented in LPS orientation. It produces the segmented brain, along with a .csv file of the calculated volumes of each ROI.
Plese make sure to use post-skull stripped T1 brain images (.nii.gz) as inputs or use NiChart_DLMUSE for the full processing pipeline from RAW T1.
pip install DLMUSEgit clone https://github.com/CBICA/DLMUSE
cd DLMUSE
pip install -e .Depending on your system configuration and supported CUDA version, you may need to follow the PyTorch Installation Instructions.
A pre-trained nnUNet model can be found at our hugging face account. Feel free to use it under the package's license.
DLMUSE -i "input_folder" -o "output_folder" -device cpufrom DLMUSE import run_dlmuse_pipeline
...
run_dlmuse_pipeline(in_dir, out_dir, device)For more details, please refer to
DLMUSE -hOur model download process creates several deep directory structures. If you are on Windows and your model download process fails, it may be due to Windows file path limitations.
To enable long path support in Windows 10, version 1607, and later, the registry key HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\FileSystem LongPathsEnabled (Type: REG_DWORD) must exist and be set to 1.
If this affects you, we recommend re-running DLMUSE with the --clear_cache flag set on the first run.
@article{doi:10.1148/ryai.240299, author = {Bashyam, Vishnu M. and Erus, Guray and Cui, Yuhan and Wu, Di and Hwang, Gyujoon and Getka, Alexander and Singh, Ashish and Aidinis, George and Baik, Kyunglok and Melhem, Randa and Mamourian, Elizabeth and Doshi, Jimit and Davison, Ashwini and Nasrallah, Ilya M. and Davatzikos, Christos and }, title = {DLMUSE: Robust Brain Segmentation in Seconds Using Deep Learning}, journal = {Radiology: Artificial Intelligence}, volume = {0}, number = {ja}, pages = {e240299}, year = {0}, doi = {10.1148/ryai.240299}, note ={PMID: 40960397}, URL = {https://doi.org/10.1148/ryai.240299}, eprint = {https://doi.org/10.1148/ryai.240299}, abstract = { “Just Accepted” papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review before it is published in its final version. Please note that during production of the final copyedited article, errors may be discovered which could affect the content. Purpose To introduce an open-source deep learning brain segmentation model for fully automated brain MRI segmentation, enabling rapid segmentation and facilitating large-scale neuroimaging research. Materials and Methods In this retrospective study, a deep learning model was developed using a diverse training dataset of 1900 MRI scans (ages 24–93 with a mean of 65 years (SD: 11.5 years) and 1007 females and 893 males) with reference labels generated using a multiatlas segmentation method with human supervision. The final model was validated using 71391 scans from 14 studies. Segmentation quality was assessed using Dice similarity and Pearson correlation coefficients with reference segmentations. Downstream predictive performance for brain age and Alzheimer’s disease was evaluated by fitting machine learning models. Statistical significance was assessed using Mann–Whittney U and McNemar’s tests. Results The DLMUSE model achieved high correlation (r = 0.93–0.95) and agreement (median Dice scores = 0.84–0.89) with reference segmentations across the testing dataset. Prediction of brain age using DLMUSE features achieved a mean absolute error of 5.08 years, similar to that of the reference method (5.15 years, P = .56). Classification of Alzheimer’s disease using DLMUSE features achieved an accuracy of 89% and F1-score of 0.80, which were comparable to values achieved by the reference method (89% and 0.79, respectively). DLMUSE segmentation speed was over 10000 times faster than that of the reference method (3.5 seconds vs 14 hours). Conclusion DLMUSE enabled rapid brain MRI segmentation, with performance comparable to that of state-of-theart methods across diverse datasets. The resulting open-source tools and user-friendly web interface can facilitate large-scale neuroimaging research and wide utilization of advanced segmentation methods. ©RSNA, 2025 } }
For more information, please contact CBICA Software.
Contributions are welcome! Please refer to our CONTRIBUTING.md for more information on how to report bugs, suggest enhancements, and contribute code. Please make sure to write tests for new code and run them before submitting a pull request.