Skin-Project

This repository contains the code for all the experiments performed in the paper: Machine Learning and Deep Learning Approaches for Classifying Keloid Images in the Context of Malignant and Benign Skin Disorders.

About

The classical approach to diagnosing various skin disorders, including keloids, relies on dermatoscopy. However, this method is often considered complex and expensive. In this study, we propose a deep learning model based on transfer learning to identify non-dermatoscopic (clinical) images of keloids among other benign and malignant skin lesions.

Setup for Running Locally

To run this project locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/OEAdebayo/skin-project.git

2. Ensure Python 3.12 or later is installed: Check your Python version:

   python -V

   If the version is below 3.12, follow this guide to install Python 3.12 using pyenv.

3. Navigate to the project root and create a virtual environment:

   cd skin-project
   python3 -m venv .venv

4. Activate the virtual environment:

   source .venv/bin/activate  # On macOS/Linux
   .venv\Scripts\activate    # On Windows

5. Install dependencies and set up required directories:

   make setup

Usage

To train the model and run experiments, use the provided script run_training.py with the appropriate arguments.

Example: Training the VGG16 Model

To train the default (i.e., VGG16) model without fine-tuning and using the original training dataset for binary classification over 10 epochs, run:

python run_training.py --classification_type bc --class_balance_type none

The results will be saved in the output directory.

Viewing Available Arguments

To see all available options for run_training.py, run:

python run_training.py --help

Citation

If you use this work in your project, please cite the following paper:

@article{OLUSEGUNETAL2025,
    title = {Deep Learning Approaches for the Classification of Keloid Images in the Context of Malignant and Benign Skin Disorders},
    author = {Olusegun Ekundayo Adebayo and Brice Chatelain and Dumitru Trucu and Raluca Eftimie},
    journal = {Diagnostics},
    volume = {15},
    year = {2025},
    issue = {6},
    pages = {710},
    doi = {10.3390/diagnostics15060710},
    url = {https://www.mdpi.com/2075-4418/15/6/710}
}

License

This project is released under the Creative Commons Attribution 4.0 International License.

---

For any issues or contributions, please feel free to open an issue or submit a pull request.