Next-Gen Image Shielding: Deep Learning’s Stealthy Image Encryption for Health-Tech IoT
Archana S. Nadhan, Israel Jeena Jacob · 2023
The healthcare sector has undergone a transformation with the advancement of the Internet of Things, facilitating the seamless incorporation of medical imaging devices within the industry's data systems, leading to faster diagnostics and treatments. Maintaining the privacy of patients is crucial and complex, especially when medical images can reveal intimate and private details about a patient's medical history. We explore the usage of advanced Deep Learning methods for encrypting medical diagnostic images, leveraging a cryptographic-based image encryption and decryption framework. The Proposed approach uses a simple key learning model built on ResNet-50 architecture to convert one visual format into another. This model specifically incorporates a domain known as "hidden factors" during the encryption phase. To revert the encrypted image to its original form, we employ reconstructive networks. By extracting relevant data from the encrypted content and enabling direct data mining from the personal data ecosystem, we offer a system that ensures a good return on investment. Additionally, the suggested method incorporates the Enhanced Encryption technique using a stream cipher. We validated the new approach using two publicly available datasets. Preliminary evaluations suggest that the method provides superior security and impressive operational efficiency.