🌟 DermaScan

DermaScan-AI is an ML-powered web application that analyzes skin lesion images to detect the likelihood of benign or malignant conditions. It also offers personalized tips and recommendations based on the prediction using integrated LLM (Large Language Model) advice.

---

🚀 Features

• 🗃 Upload skin lesion images
• 🤖 TensorFlow.js model (client-side) for instant prediction
• 📊 Result with confidence score (Benign / Malignant)
• 🤠 LLM integration for actionable advice (sunscreen tips, precautions, etc.)
• 🔐 Secure login/signup using JWT + MongoDB
• 💡 Responsive & user-friendly UI built with React + Tailwind

---

📁 Project Structure

dermaScan/
├── frontend/         # React-based UI
│   ├── src/
│   │   ├── App.jsx
│   │   ├── AppDashboard.jsx
│   │   └── Registration.jsx
│   └── public/
│       └── model/    # TensorFlow.js model files
├── skin-cancer-backend/
│   ├── server.js     # Express server
│   ├── routes/
│   │   ├── auth.js
│   │   └── advice.js
│   ├── models/
│   │   └── user.js
│   └── .env

---

---

🛠 Tech Stack

• Frontend: React, Tailwind CSS, TensorFlow.js
• Backend: Node.js, Express, MongoDB
• Authentication: JWT
• AI Model DermaScan-AI uses a custom-trained Convolutional Neural Network (CNN) built with TensorFlow.js to analyze skin lesion images directly in the browser. The model classifies uploaded images into two categories:

Malignant (possibly cancerous)

Benign (non-cancerous)

Key features:

Runs fully in the browser — ensuring user privacy as no images are uploaded to any server.

Delivers real-time predictions with a clear confidence percentage.

Built using TensorFlow.js for smooth and responsive web integration.

• LLM Integration: OpenAI / Gemini API (custom advice based on result)

---

---

⚙️ Setup Instructions

1. Clone the repository

git clone https://github.com/Deepak5106/DermaScan.git
cd DermaScan

---

2. Setup Backend

cd skin-cancer-backend
npm install

Create .env file

PORT=5000
MONGO_URI=mongodb://127.0.0.1:27017/dermascan
JWT_SECRET=your_super_secret_key
OPENAI_API_KEY=your_openai_key

Start the server

node server.js

---

3. Setup Frontend

cd frontend
npm install
npm run dev

---

💡 How It Works

1. User logs in or signs up

2. Uploads an image of a skin lesion

3. TensorFlow.js model analyzes the image locally

4. Result is shown with confidence (e.g., Malignant (92.5%))

5. Based on result:

   • Benign → sunscreen, skincare tips
   • Malignant → consult dermatologist, early symptoms advice

6. LLM gives context-aware tips via API response

---

🔐 Authentication

• Uses JWT to protect routes
• MongoDB stores user credentials securely (hashed via bcrypt)

---

✨ Future Improvements

• 🦬 Multi-class prediction (e.g., melanoma, carcinoma)
• 📱 PWA support for mobile
• 🌍 Upload image from camera
• 📉 View historical scans and progress over time
• 🔣 Voice assistant for skin care tips

---

🧑‍💻 Developed By

Deepak – ECE Student Feel free to connect on LinkedIn or contribute to the repo!