🧠 DCT-Based Medical Report Steganography API

This project implements an API-driven, secure and error-proof medical report handling system using DCT-based image steganography. It is specifically designed to embed patient IDs into medical images (like X-rays, MRI scans, etc.) for automated verification, ensuring privacy, accuracy, and tamper-proof storage in hospital databases.

---

📌 Table of Contents

• 🧠 DCT-Based Medical Report Steganography API
  • 📌 Table of Contents
  • 🎯 Problem Statement
  • 💡 Solution Overview
  • 🛠 Features
  • 🚀 How to Run
    • 🔧 Step 1: Create Virtual Environment
    • 📦 Step 2: Install Dependencies
    • ▶️ Step 3: Run API
  • 📤 API Endpoints
    • ➕ /embed [POST]
    • 🔍 /extract [POST]
  • 📁 Directory Structure
  • 📦 Dependencies
  • 🔐 Security & Use Case
  • 📌 Future Enhancements
  • 🧑‍💻 Author

---

🎯 Problem Statement

Hospitals currently rely on manual entry or non-validated automation for storing patient medical reports. This can result in:

• Human errors (wrong report mapped to wrong patient)
• No security validation between stored image and patient ID
• No privacy control in case of unauthorized report viewing

---

💡 Solution Overview

✅ This project uses DCT (Discrete Cosine Transform) and adaptive steganography to:

• Embed the patient ID securely into a medical image
• Automate storage of these verified stego-images
• Allow the doctor to extract the embedded ID for cross-validation before report viewing

The solution ensures zero human dependency, no data mix-up, and enhanced patient confidentiality.

---

🛠 Features

• 🔒 DCT-based adaptive image steganography
• 🧠 Texture and perceptual masking for imperceptibility
• 📷 Embed any patient ID into any image (X-ray, MRI, CT)
• ✅ REST API with /embed and /extract endpoints
• 🌐 No need for image to be in any specific folder
• 🔁 Support for image uploads and dynamic processing
• 💾 Stego-image returned automatically after upload
• 📜 Extraction matches embedded ID with ground truth

---

🚀 How to Run

🔧 Step 1: Create Virtual Environment

python3 -m venv venv
source venv/bin/activate  # macOS/Linux
venv\Scripts\activate     # Windows

📦 Step 2: Install Dependencies

pip install -r requirements.txt

▶️ Step 3: Run API

uvicorn dct_fast_api:app --reload

---

📤 API Endpoints

➕ /embed [POST]

Uploads an image and patient ID, returns a stego-image.

Form Data:

• file: image file (JPG/PNG)
• patient_id: patient ID string

Returns:
Stego-image with the embedded patient ID.

---

🔍 /extract [POST]

Uploads a stego-image and extracts the embedded patient ID.

Form Data:

• file: stego-image

Returns:
Extracted patient ID string.

---

📁 Directory Structure

.
├── dct_fast_api.py         # FastAPI app
├── dct_claude_quant.py     # Embedding & extraction logic
├── README.md               # This file
├── requirements.txt
├── Embedded-image/         # Output images
├── Decoded-message/        # Extracted text
├── Images/                 # Input images (optional)
├── screenshots/            # Screenshots for demonstration

---

📦 Dependencies

.
├── fastapi 
├── uvicorn     
├── opencv-python            
├── numpy
├── pillow       
├── scipy       
├── python-multipart 

** Installing dependencies **

pip install -r requirements.txt

---

🔐 Security & Use Case

• Ensures privacy-preserving storage of medical images
• Prevents unauthorized report access
• Protects against mismatched patient-report mapping
• Automates the workflow of hospital database entry

---

📌 Future Enhancements

• 🔒 AES encryption of patient ID before embedding
• 🖥️ GUI for hospital-side integration
• 🔌 Plugin for PACS/RIS systems
• 📜 Embed medical report hash + timestamp for audit trail
• 📦 Embed multiple fields (name, ID, report date) in compressed form

---

🧑‍💻 Author

Saniel Bhattarai
B.Tech in Computer Science Engineering
Specialization: Information Security Final Year Capstone Project (2025)

Riya Vaid
B.Tech in Computer Science Engineering
Specialization: Information Security
VIT Vellore
Final Year Capstone Project (2025)