🧠 BRISC 2025: Brain Tumor MRI Dataset for Segmentation and Classification

• Published in: Scientific Data (Nature Portfolio)
  DOI: https://doi.org/10.1038/s41597-026-06753-y

BRISC is a high-quality, expert-annotated MRI dataset curated for brain tumor segmentation and classification. It addresses common limitations in existing datasets (e.g., BraTS, Figshare), including class imbalance, narrow tumor focus, and annotation inconsistencies.

This dataset includes:

• 6,000 T1-weighted MRI images
• Four classes: Glioma, Meningioma, Pituitary Tumor, and No Tumor
• Pixel-wise segmentation masks validated by physicians and radiologists
• Three anatomical planes: Axial, Coronal, and Sagittal
• Clean, stratified training (5,000) and testing (1,000) splits

📦 Dataset Contents

The BRISC dataset is organized into two main tasks:

1. Classification Task — for multi-class tumor identification
2. Segmentation Task — for pixel-wise tumor region annotation

📁 Folder Descriptions

classification_task/

• Contains subfolders organized by tumor class (glioma, meningioma, pituitary, no_tumor)
• Each class folder includes raw T1-weighted MRI slices in .jpg format
• Supports training of image-level classification models

segmentation_task/

• Contains paired MRI images/ and corresponding binary masks/
• All slices are from T1-weighted MRI scans
• Masks are pixel-level annotations created and verified by medical experts
• Image and mask filenames are aligned

📦 Dataset structure

brisc2025/
├─ classification_task/
│  ├─ train/
│  │  ├─ glioma/
│  │  │  ├─ brisc2025_train_00001_gl_ax_t1.jpg
│  │  │  └─ ...
│  │  ├─ meningioma/
│  │  ├─ pituitary/
│  │  └─ no_tumor/
│  └─ test/
│     ├─ glioma/
│     │  ├─ brisc2025_test_00001_gl_ax_t1.jpg
│     │  └─ ...
│     ├─ meningioma/
│     ├─ pituitary/
│     └─ no_tumor/
├─ segmentation_task/
│  ├─ train/
│  │  ├─ images/
│  │  │  ├─ brisc2025_train_00001_gl_ax_t1.jpg
│  │  │  └─ ...
│  │  └─ masks/
│  │     ├─ brisc2025_train_00001_gl_ax_t1.png
│  │     └─ ...
│  └─ test/
│     ├─ images/
│     │  ├─ brisc2025_test_00001_gl_ax_t1.jpg
│     │  └─ ...
│     └─ masks/
│        ├─ brisc2025_test_00001_gl_ax_t1.png
│        └─ ...
├─ manifest.json
├─ manifest.csv
├─ manifest.json.sha256
├─ manifest.csv.sha256
└─ README.md

Notes:

• Classification folders contain image-level labels suitable for standard image classification pipelines.
• Segmentation folders contain paired MRI images/ and corresponding binary masks/.
• Image and mask filenames are identical except for file extension (images: .jpg, masks: .png).
• All images are T1-weighted slices.

🧪 Technical Details

• Total samples: 6,000 (5,000 training / 1,000 testing)
• Annotation quality: Reviewed and corrected by medical experts
• Imaging modality: Only T1-weighted MRI
• Planes: Balanced representation across axial, coronal, and sagittal
• Tumor classes: Balanced distribution across four categories

• Example Filename: brisc2025_test_00010_gl_ax_t1.jpg

🎯 Why Use BRISC?

• ✅ Balanced and diverse multi-class labels
• ✅ Expert-refined segmentation masks with medical consensus
• ✅ Includes challenging real-world cases and subtle tumors
• ✅ Designed for multi-task learning: segmentation + classification
• ✅ Ideal for training robust models with real-world MRI variability

📈 Applications

• Brain tumor segmentation using deep learning
• Multi-class classification and binary tumor detection
• Cross-plane generalization in 2D medical imaging
• Developing AI-assisted diagnosis tools in neuro-oncology

📄 Citation & Publication

This dataset is introduced in our publication:
"BRISC: Annotated Dataset for Brain Tumor Segmentation and Classification with Swin-HAFNet" Fateh et al., 2025

If you use the BRISC dataset in your research, please cite our paper:

###BibTeX

@article {fateh2025brisc,
  title={Brisc: Annotated dataset for brain tumor segmentation and classification with swin-hafnet},
  author={Fateh, Amirreza and Rezvani, Yasin and Moayedi, Sara and Rezvani, Sadjad and Fateh, Fatemeh and Fateh, Mansoor and Abolghasemi, Vahid},
  journal={arXiv preprint arXiv:2506.14318},
  year={2025}
}

🔗 Acknowledgments

• Credits: Special thanks to our collaborating radiologists and physicians for expert annotation and review.

📚 Sources

The BRISC dataset was developed with reference to and inspiration from several widely used and cited brain tumor imaging datasets. These include:

• Masoud Nick Parvar Brain Tumor MRI Dataset
• Sartaj Bhuvaji Brain Tumor Classification MRI Dataset
• Cheng’s Brain Tumor Dataset
• Ahmed Hamada Brain Tumor Detection Dataset
• ESFIAM Brain Tumor MRI Dataset
• Pradeep2665 Brain MRI Dataset
• Mohamed Metwaly Sherif Brain Tumor Dataset
• Nyalkakadia MR Images of Brain for Tumor Classification Dataset
• Brima's Brain MRI Dataset
• IXI Dataset