MedSecure: Watermarking Technique for Medical Images for Authentication and Copyright Protection

Dr Archana Tiwari, Sudev Kumar Padhi, Umesh Kashyap, SK Subidh Ali

Keywords: Deep Learning, Watermarking, Copyright Protection, Authentication, Telemedicine.

TL;DR: Watermarking Technique for Medical Images for Authentication and Copyright Protection

Abstract:

The surge in telemedicine adoption has underscored the critical need for secure medical image transmission and storage. However, existing techniques struggle to balance imperceptibility, resilience to acceptable image manipulations, and robustness against adversarial threats. We propose a deep learning-based dual watermarking framework that embeds a perceptual hash for copyright protection and a cryptographic hash for integrity verification. By incorporating deep learning, our method ensures robustness against surrogate model and content-preserving attacks while preserving diagnostic fidelity. The experimental results demonstrate imperceptibility (PSNR: 40.23 dB, SSIM: 0.98) and with an accuracy of 95.4% against adversarial manipulations, which set a new benchmark for secure medical
image authentication in telemedicine.

Submission Number: 51