Dermatology Assistant Web Application

A web-based dermatology assistant that combines image classification and natural language processing to provide medical insights and answers to user queries.

Features

• User authentication (login/register)
• Image upload and classification
• Chat interface with message history
• Integration with ConvNeXt-Tiny classifier
• Integration with Mistral-7B-Instruct LLM
• Retrieval-Augmented Generation (RAG) for enhanced responses

Prerequisites

• Python 3.8+
• PyTorch
• Flask
• SQLite (or PostgreSQL)
• Your trained ConvNeXt-Tiny model
• Your Mistral-7B-Instruct model
• Your RAG implementation

Installation

1. Clone the repository:

git clone <repository-url>
cd dermatology-assistant

2. Create a virtual environment:

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

3. Install dependencies:

pip install -r requirements.txt

4. Set up environment variables:

export SECRET_KEY="your-secret-key-here"  # On Windows: set SECRET_KEY=your-secret-key-here

5. Initialize the database:

flask db init
flask db migrate
flask db upgrade

Configuration

1. Update the model paths in models.py:

   • Set the path to your trained ConvNeXt-Tiny model
   • Set the path to your Mistral-7B-Instruct model
   • Configure your RAG implementation

2. (Optional) Configure PostgreSQL:

   • Update SQLALCHEMY_DATABASE_URI in app.py with your PostgreSQL connection string

Running the Application

1. Start the Flask development server:

python app.py

2. Access the application at http://localhost:5000

Usage

1. Register a new account or log in
2. Start a new chat
3. Upload a dermatology image with your first message
4. Continue the conversation with text-only messages
5. View chat history in the left panel

Project Structure

dermatology-assistant/
├── app.py              # Main Flask application
├── models.py           # Model integration (classifier, LLM, RAG)
├── requirements.txt    # Python dependencies
├── templates/          # HTML templates
│   ├── base.html
│   ├── login.html
│   ├── register.html
│   └── chat.html
├── static/            # Static files (CSS, JS)
├── uploads/           # Uploaded images
└── instance/          # Database files

Security Considerations

• All user passwords are hashed using Werkzeug's security functions
• File uploads are restricted to image files
• Maximum file size is limited to 16MB
• User authentication is required for all sensitive routes

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

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