princedex/brain-tumor-detection-system

NeuroScan AI - Brain Tumor Detection System

An AI-powered brain tumor detection and classification system built using Deep Learning and Medical Imaging as a Final Year Project.

Overview

This system classifies brain MRI scans into four categories:

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

Model

• Architecture: EfficientNetV2-B0 with Transfer Learning
• Strategy: Two-phase training (head training + fine-tuning)
• Explainability: Grad-CAM heatmap visualization
• Parameters: ~7.1 Million
• Framework: TensorFlow 2.x / Keras

Results

Metric	Value
Test Accuracy	71.57%
Validation Accuracy	82.20%
AUC-ROC	0.8928
No Tumor Sensitivity	90.48%
Glioma Sensitivity	39.00%

Model Comparison

Metric	EfficientNetV2	ResNet50
Test Accuracy	71.57%	76.90%
Val Accuracy	82.20%	89.70%
AUC-ROC	0.8928	0.9392
Parameters	~7.1M	~25.6M
Glioma Sensitivity	39.00%	29.00%

How to Run

1. Clone the repository

git clone https://github.com/YOUR_USERNAME/neuroscan-ai.git
cd neuroscan-ai

2. Install dependencies

pip install -r requirements.txt

3. Download the dataset

Download the Kaggle Brain MRI Dataset from: https://www.kaggle.com/datasets/sartajbhuvaji/brain-tumor-classification-mri and place it in the archive/ folder.

4. Run the web app

streamlit run streamlit_app.py

Project Structure

neuroscan-ai/
├── streamlit_app.py           # Web application
├── BTP.ipynb                  # Training notebook
├── best_model_v2_phase2.keras # Trained model weights
├── requirements.txt           # Dependencies
├── static/uploads/            # Temporary upload folder
└── archive/                   # Dataset (not included)
    ├── Training/
    └── Testing/

Features

• Multi-class brain tumor classification
• Grad-CAM visual explainability
• Per-class Sensitivity & Specificity reporting
• EfficientNetV2 vs ResNet50 comparison
• Interactive Streamlit web interface

Disclaimer

This system is a research prototype for academic purposes only. It is not intended for clinical use and must not replace professional medical diagnosis.

Developer

• Name: Stephen Ifeanyi
• Brand: Princedex
• Email: ifeanyistephen003@gmail.com

Dataset

Chakrabarty, N. (2021). Brain MRI Images for Brain Tumor Detection. Kaggle. https://www.kaggle.com/datasets/sartajbhuvaji/brain-tumor-classification-mri