aparida12/brats2025 repository overview

⁠Neural Network-Based Brain MRI Segmentation for BraTS 2025 Submissions

This repository contains the inference engine of Pediatric Brain MRI Segmentation for BraTS 2025 submitted by the Children's National Hospital Team. The inference engine is Dockerized to allow for a plug-and-play approach for segmentation of NifTi data.

⁠Easily segment your data with one command

Once all the Docker and its dependencies are installed, you can simply test the pediatric brain tumor segmentation:

Run with GPUs:

default:

docker run --rm --network none --gpus=all --memory=16G --shm-size 4G \
        -v <inpdir>:/input/:ro -v <outdir>:/output/:rw \
        aparida12/brats2025:<tasktag>

where:

• <inpdir> is the directory with all input MRI NifTi files(extn .nii.gz; .nii will be ignored) that need to be segmented.
• <outdir> is the directory where all outputs are stored(<inpdir> should not be <outdir>).
• <tasktag> is the tag corresponding to the BraTS 2025 challenge

⁠BraTS 2025 Tasks

The repo contains top-performing segmentation algorithms for the challenges of Adult Glioma Pre and Post Treatment (GLI), Pediatric Tumor (PED), Adult Glioma in Sub-Saharan African population (SSA), Adult Metastasis (MET), Adult Meningioma (MEN), and Meningioma Radiotherapy(MENRT). The table displays the tags of the various Docker containers, corresponding to their respective tasks.

⁠Input Folder Structure

The input folder should follow a similar structure as shown below, including the naming convention do not add any other files or folders into the <inpdir>. XX can be anything. Ensure the image has been preprocessed using the same steps followed in BraTS-PEDs preprocessing for 2024 data.

<inpdir>
│
└───XX-XX-XX-XX
│   │   XX-XX-XX-XX-t1n.nii.gz
│   │   XX-XX-XX-XX-t1c.nii.gz
│   │   XX-XX-XX-XX-t2w.nii.gz
│   │   XX-XX-XX-XX-t2f.nii.gz
│    |.

⁠Singularity container

Once all the singularity and its dependencies are installed, you can test the pediatric brain tumor segmentation:

singularity pull brats-peds-2024.sif docker://aparida12/brats-peds-2024:v20240913
singularity build --sandbox brats-peds-2024 docker://aparida12/brats-peds-2024:v20240913

mkdir brats-peds-2024/input/
mkdir brats-peds-2024/output/

singularity exec --writable --nv --bind  <inpdir>:/input/ --bind <outdir>:/output brats-peds-2024 bash -c “export MKL_THREADING_LAYER=GNU;cd /mlcube_project;python3 mlcube.py infer --data_path=/input --output_path=/output"

⁠Citations

If you use and/or refer to this software in your research, please cite the following papers:

• A. Parida, D. Capellán-Martín, Z. Jiang, N. Kulkarni, K. Iyer, A. Tapp, S. M. Anwar, M. J. Ledesma-Carbayo, M. G. Linguraru. Improving Pre-trained Adult Glioma Segmentation Models Using only Post-processing Techniques. In: Bakas, S., et al. Segmentation, Classification, and Synthesis for Brain Tumors and Traumatic Brain Injuries. MICCAI 2025. Lecture Notes in Computer Science, vol 16376. Springer, Cham. https://doi.org/10.1007/978-3-032-16365-3_22⁠

• D. Capellán-Martín, A. Parida, Z. Jiang, N. Kulkarni, K. Iyer, A. Tapp, S. M. Anwar, M. J. Ledesma-Carbayo, M. G. Linguraru. Adaptable Segmentation Pipeline for Diverse Brain Tumors with Radiomic-Guided Subtyping and Lesion-Wise Model Ensemble. In: Bakas, S., et al. Segmentation, Classification, and Synthesis for Brain Tumors and Traumatic Brain Injuries. MICCAI 2025. Lecture Notes in Computer Science, vol 16376. Springer, Cham. https://doi.org/10.1007/978-3-032-16365-3_41⁠

• D. Capellán-Martín, Z. Jiang, A. Parida, X. Liu, V. Lam, H. Nisar, A. Tapp, S. Elsharkawi, M. J. Ledesma-Carbayo, S. M. Anwar, M. G. Linguraru. Model Ensemble for Brain Tumor Segmentation in Magnetic Resonance Imaging. In: Baid, U., et al. Brain Tumor Segmentation, and Cross-Modality Domain Adaptation for Medical Image Segmentation. crossMoDA BraTS 2023 2023. Lecture Notes in Computer Science, vol 14669. Springer, Cham. https://doi.org/10.1007/978-3-031-76163-8_20⁠

• Z. Jiang, D. Capellán-Martín, A. Parida, X. Liu, M. J. Ledesma-Carbayo, S. M. Anwar, M. G. Linguraru, "Enhancing Generalizability in Brain Tumor Segmentation: Model Ensemble with Adaptive Post-Processing,"  2024 IEEE International Symposium on Biomedical Imaging (ISBI), Athens, Greece, 2024, pp. 1-4, doi: 10.1109/ISBI56570.2024.10635469⁠.

⁠Disclaimer

This software is provided without any warranties or liabilities and is intended for research purposes only. It has not been reviewed or approved for clinical use by the Food and Drug Administration (FDA) or any other federal or state agency.

⁠External Links

• Code Repo⁠
• Interactive Webapp⁠