Brain Tumor Classification

Overview 🌟

This project was carried out as part of a statistical learning course. It is a first project in image processing.

The goal of this project is to detect and classify brain tumors using MRI images. It supports four classes:

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

The project includes features for:

• Data exploration
• Model evaluation metrics
• Explainability analyses
• A Streamlit application to test the model

---

Dataset 📊

The dataset used for this project is available on Kaggle: Brain Tumor Classification (MRI).

Data Summary

• MRIs: Classified into the four categories listed above.
• Folder structure: Separate folders for training, validation, and testing.
• Preprocessing: Images are resized, normalized, and converted to grayscale for model training.

---

Features 🛠️

1. Data Exploration 🔍

An interactive HTML report provides a complete overview of the dataset.

2. Model Metrics 📈

A detailed HTML report with evaluation metrics such as accuracy, recall, and F1-score.

3. Explainability 🧠

Visual analyses of the model’s decisions using Xplique.

4. Streamlit Application 🌐

An intuitive interface that allows users to upload MRI images and obtain real-time classification results.

Note: The application requires the pre-trained model, which is not included in this GitHub repository. You must place the model file in the appropriate folder before running the application.

---

Installation and Execution ⚙️

Prerequisites

• Python: Version 3.11 or higher
• System requirements: A system capable of running virtual environments

Instructions

1. Clone the repository:

   git clone <repository-url>
   cd <repository-folder>

2. Run the installation script (Linux/Mac):

   ./lancez_moi.sh

   This script will perform the following actions:

• Create a virtual environment.
• Activate the virtual environment.
• Open the data exploration report in your default browser.
• Display the model metrics report.
• Access the explainability analyses.
• Launch the Streamlit application.

3. Run the installation script (Windows):

   lancez_moi_windows.sh

   This script is designed for Windows users.

4. Run the Streamlit application manually:

   cd src/models
   streamlit run CNN_app.py

---

Folder Structure 📂

project-root/
|
|-- data/                 # Folder containing the dataset (train, validation, test)
|-- Data_visualisation/   # Scripts and files for data visualization
|-- results/              # Reports and results (e.g., metrics.html, explainability analyses)
|-- src/                  # Project source code
|   |-- models/           # Scripts for models, training, and detection
|   |   |-- CNN_app.py    # Streamlit application for classification
|   |   |-- CNN_train.py  # Script to train the CNN model
|   |   |-- preprocessing/ # Image preprocessing modules
|-- venv/                 # Virtual environment files
|-- .gitignore            # Git ignore rules
|-- config.yml            # Project configuration file
|-- lancez_moi.sh         # Installation script for Linux/Mac
|-- lancez_moi_windows.sh # Installation script for Windows
|-- README.md             # Project documentation
|-- requirements.txt      # Python dependencies

---

README translated from the french version written by Lilya-Nada KHELID @Lilyakhelid.