AML Detection using Deep Learning

A deep learning project for detecting Acute Myeloid Leukemia (AML) from blood cell images using Convolutional Neural Networks and Transfer Learning.

📋 Project Overview

This project implements a two-phase approach for AML detection:

• Phase 1: Basic CNN model for initial classification
• Phase 2: Transfer Learning with EfficientNetB0 for improved accuracy

🗂️ Project Structure

aml-detection/
├── data/                          # Dataset (see data/README.md for download)
│   ├── Blood cell Cancer [ALL]/
│   └── raw/
├── models/                        # Trained models
│   ├── phase1/
│   │   ├── basic_cnn.h5
│   │   └── best_model.h5
│   └── phase2/
│       ├── best_efficientnetb0.h5
│       └── efficientnetb0_final.h5
├── results/                       # Training results and visualizations
├── scripts/                       # Training scripts
│   ├── phase1_basic_model.py
│   └── phase2_transfer_learning.py
└── requirements.txt               # Python dependencies

📊 Dataset

Note: This project uses the Blood Cell Cancer [ALL] - 4 Class dataset from Kaggle.

Download Instructions:

📥 Kaggle Dataset: Blood Cell Cancer [ALL] - 4 Class

• Size: ~6 GB
• Format: Blood cell microscopy images
• Classes: 4 classes (Malignant early Pre-B, Pre-B, Pro-B, and Benign)

Setup the Dataset:

Method 1: Manual Download

1. Visit the Kaggle dataset page
2. Click "Download" (requires free Kaggle account)
3. Extract the downloaded zip file
4. Move the extracted folder to the data/ directory

Method 2: Using Kaggle API

# Install Kaggle API
pip install kaggle

# Download dataset (make sure your Kaggle API credentials are configured)
kaggle datasets download -d mohammadamireshraghi/blood-cell-cancer-all-4class

# Extract to data folder
unzip blood-cell-cancer-all-4class.zip -d data/

For more details, see data/README.md

🚀 Getting Started

Prerequisites

• Python 3.8+
• TensorFlow 2.x
• CUDA-compatible GPU (recommended)

Installation

1. Clone the repository:

git clone https://github.com/Parisaroozgarian/aml-detection.git
cd aml-detection

2. Create virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Download the dataset (see Dataset section above)

💻 Usage

Phase 1 - Basic CNN:

python scripts/phase1_basic_model.py

Phase 2 - Transfer Learning:

python scripts/phase2_transfer_learning.py

📈 Results

Training results, visualizations, and performance metrics are available in the results/ folder.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📝 License

This project is licensed under the MIT License.

🙏 Acknowledgments

• Dataset source: Blood Cells Cancer (ALL) dataset. (2021, October 30). Kaggle. https://www.kaggle.com/datasets/mohammadamireshraghi/blood-cell-cancer-all-4class
• Inspired by medical AI research in leukemia detection

📚 Citation

If you use this dataset in your research, please cite:

Blood Cells Cancer (ALL) dataset. (2021, October 30). Kaggle. https://www.kaggle.com/datasets/mohammadamireshraghi/blood-cell-cancer-all-4class