🧠 Brain Tumor Classification using MRI Images

📌 Project Overview

Brain tumor detection is a critical medical task. Early detection can significantly improve patient survival rates.
This project uses Deep Learning (CNN & VGG16) to classify MRI brain images into four categories:

• Glioma Tumor
• Meningioma Tumor
• Pituitary Tumor
• No Tumor

The system is deployed as a web application using Flask (backend) and React (frontend).

---

🎯 Objectives

• Build a deep learning model for brain tumor classification
• Compare two models (Custom CNN vs VGG16)
• Develop a full ML pipeline
• Deploy a working web application

---

📊 Dataset

• Source: Kaggle
• Dataset Link: https://www.kaggle.com/datasets/sartajbhuvaji/brain-tumor-classification-mri

Dataset Details:

• Total Images: ~3000
• Classes:
  • Glioma (~826)
  • Meningioma (~822)
  • Pituitary (~827)
  • No Tumor (~395)

---

🧠 Features & Target

• Features: MRI Image pixel values (64 × 64 × 3)
• Target Variable: Tumor Type (4 classes)

---

⚙️ Technologies Used

• Python
• TensorFlow / Keras
• NumPy, Matplotlib, Scikit-learn
• Flask (Backend)
• React.js (Frontend)

---

🔄 Machine Learning Pipeline

1. Data Preprocessing

• Image rescaling (normalization)
• Data augmentation:
  • Shear transformation
  • Zoom
  • Horizontal flipping
• Train-Test split using directory structure

---

2. Models Used

✅ Model 1: Custom CNN

• Convolutional Layers
• MaxPooling Layers
• Fully Connected Layers

✅ Model 2: VGG16 (Transfer Learning)

• Pretrained on ImageNet
• Frozen base layers
• Custom classifier added

Features of web application:

• Upload MRI image
• Preview image before prediction
• Predict tumor type
• Color-coded result:
  • 🟢 Green → No Tumor
  • 🔴 Red → Tumor detected

---