NeuroScan: Brain Tumor Detection System

An AI-powered brain tumor detection system using machine learning and deep learning models. Includes classification with CNN, VGG16, AdaBoost, CatBoost, and a weighted ensemble method. Built with Python, trained in Google Colab, and deployed using Flask.

---

📁 Project Structure

NeuroScan.BrainTumorDetection/
├── models/                  # Trained models (CNN, VGG16, AdaBoost, CatBoost)
├── data/ # Sample dataset (limited due to size)
|
|---Notebooks/
|   |---BrainTumorDetection.ipynb
|   |---model_evaluation.ipynb
|
├── src/
│   ├── app.py              # Flask backend
│   ├── utils/              # Helper functions
│   └── ...
├── templates/              # HTML interface for web app
├── static/                 # CSS, JS, and assets
├── requirements.txt        # Python dependencies
└── README.md               # Project documentation

---

🚀 Features

• Brain tumor classification: Tumor / No Tumor
• Weighted ensemble voting system for robust predictions
• Flask web interface showing ensemble + individual model predictions
• Trained in Google Colab using GPU for fast experimentation

---

🧠 Models Used

• CNN (custom architecture)
• VGG16 (Transfer Learning)
• AdaBoostClassifier
• CatBoostClassifier (GPU-enabled)
• Weighted average ensemble

---

🛠️ Dependencies

Main libraries:

• TensorFlow / Keras
• scikit-learn
• CatBoost
• OpenCV
• Flask
• Albumentations

For a full list, see requirements.txt.

---

💾 Download

Due to the large size of this project, it's not fully included on CD. You can download the complete project from the GitHub repository:

GitHub Repository: https://github.com/josiaO/NeuroScan.BrainTumorDetection.git

To clone:

git clone https://github.com/josiaO/NeuroScan.BrainTumorDetection.git

---

🧪 How to Run

1. Create a virtual environment:

   python -m venv .venv
   source .venv/bin/activate  # Or .venv\Scripts\activate on Windows

2. Install dependencies:

   pip install -r requirements.txt

3. Run the Flask app:

   python src/app.py

---

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.

---

📬 Contact

Developer: Josiah Mosses

[email protected] Feel free to connect on GitHub or via email (if needed for your institution).

---

"Tech isn’t just code — it’s impact."