the-shoaib2/Brain_Tumor_MRI

Brain Tumor MRI Dataset

A curated dataset of brain MRI scans for tumor classification, organized into 4 categories.

Dataset Description

This dataset contains MRI brain scans classified into four categories:

• Glioma: Tumors that begin in the glial cells
• Meningioma: Tumors that form in the meninges
• Pituitary: Tumors in the pituitary gland
• No Tumor: Healthy brain scans

Dataset Structure

The dataset is split into three subsets:

• Training Set: 8444 images (70.0%)
• Validation Set: 1807 images (15.0%)
• Test Set: 1813 images (15.0%)

Total: 12064 images

Class Distribution

Training Set

• glioma: 2641 images
• meningioma: 1910 images
• notumor: 1702 images
• pituitary: 2191 images

Validation Set

• glioma: 565 images
• meningioma: 409 images
• notumor: 364 images
• pituitary: 469 images

Test Set

• glioma: 567 images
• meningioma: 410 images
• notumor: 366 images
• pituitary: 470 images

Directory Structure

dataset/
├── train/
│   ├── glioma/
│   ├── meningioma/
│   ├── notumor/
│   └── pituitary/
├── val/
│   ├── glioma/
│   ├── meningioma/
│   ├── notumor/
│   └── pituitary/
└── test/
    ├── glioma/
    ├── meningioma/
    ├── notumor/
    └── pituitary/

Usage

Using Hugging Face Datasets

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("the-shoaib2/Brain_Tumor_MRI")

# Access splits
train_data = dataset['train']
val_data = dataset['validation']
test_data = dataset['test']

Using with TensorFlow/Keras

import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator

# Download and extract dataset
# Then use ImageDataGenerator

datagen = ImageDataGenerator(rescale=1./255)

train_generator = datagen.flow_from_directory(
    'dataset/train',
    target_size=(299, 299),
    batch_size=32,
    class_mode='categorical'
)

Using with PyTorch

from torchvision import datasets, transforms
from torch.utils.data import DataLoader

transform = transforms.Compose([
    transforms.Resize((299, 299)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
])

train_dataset = datasets.ImageFolder('dataset/train', transform=transform)
train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)

Model

A pre-trained model using this dataset is available at: the-shoaib2/Brain_Tumor_MRI_Classification

Image Specifications

• Format: JPG/PNG
• Recommended Input Size: 299x299 pixels
• Color Mode: RGB
• Preprocessing: Normalization to [0, 1] range

Applications

This dataset can be used for:

• Brain tumor classification
• Medical image analysis research
• Deep learning model training and evaluation
• Transfer learning experiments
• Computer-aided diagnosis systems

Ethical Considerations

• This dataset is for research and educational purposes only
• Not intended for clinical diagnosis without proper medical validation
• Should be used in conjunction with expert medical opinion
• Privacy: All images are anonymized

Citation

If you use this dataset, please cite:

@misc{brain_tumor_mri_dataset,
  author = {Shoaib},
  title = {Brain Tumor MRI Dataset},
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/the-shoaib2/Brain_Tumor_MRI}}
}

License

MIT License - See LICENSE file for details

Acknowledgments

This dataset has been curated and organized for machine learning applications in medical imaging.