Brain Tumor Classification

This project showcases a deep learning-based web application for classifying brain tumors using MRI scans. The app is built with Gradio and deployed on Hugging Face Spaces.

🌟 Features

• Input: Upload an MRI image of the brain.

• Output: Predicts one of the following classes:

  • Glioma

  • Meningioma

  • Pituitary Tumor

  • Healthy (No Tumor)

• Built using:

  • Transfer Learning with models like ResNet-18 and Vision Transformers.

  • Gradio for the interactive interface.

🚀 Technology Stack

• Framework: PyTorch

• Deployment: Hugging Face Spaces

• Interface: Gradio

📊 Dataset

The app uses MRI scans from the Kaggle Brain Tumor Dataset, preprocessed and augmented for better generalization.

🔗 Live App

Try the app here: Brain Tumor Classification App

🧠 Model Details

The app uses a pretrained Vision Transformer fine-tuned on a brain tumor classification task. Key features:

• Transfer learning from ImageNet weights.

• Final classification head replaced to match the 4 output classes.

📚 How It Works

1. Upload an MRI scan (JPEG/PNG).

2. The app processes the image and predicts the class with associated probabilities.