REYHealth: Confidentiality-Preserving electronic health records sharing using adaptive N-Image steganography and huffman coding

Abstract

• Working as a reversible two-stage steganographic framework for medical images. • Protecting electronic health record patient utilizing medical images • Proposing a hierarchical, tree-structured partitioning algorithm to distribute payloads across multiple stego images. • Essential for high-resolution medical image exchange with minimal perceptual impact. • Reducing the number of bits to embed, minimizing pixel modifications and preserving visual quality. • Experimental results on medical images show the average PSNR is above 60 dB, SSIM is close to 0.99, BPP exceeds 0.4. In past few years, the digital transmission technology has rapidly evolved. One major developments is the sharing of Electronic Health Records (EHRs) in telemedicine, increasing the risk of privacy leaks. Therefore, ensuring security while preserving image quality is essential. While steganography enables embedding sensitive data, many existing methods introduce distortion or lack scalability, compromising diagnostic quality. This research introduces REYHealth, a novel adaptive framework designed to confidentially and reversibly conceal EHRs with minimal distortion and high scalability. The primary novelty lies in distributing payload data across multiple stego images, enhancing embedding capacity. The method employs a two-stage pipeline: (1) lossless Group-Average Based Difference Expansion to embed control data and part of the payload, and (2) a novel tree-structured algorithm allocating the remaining payload across multiple images based on capacity and imperceptibility. Additionally, Huffman coding compresses payload data before embedding to reduce distortion. Experimentation shows that REYHealth achieves high imperceptibility, with an average Peak Signal-to-Noise Ratio (PSNR) above 60 decibels (dB) and a Structural Similarity Index Measure (SSIM) close to 0.99. Regarding the security, the framework has vast combinatorial complexity, making it robust against steganalysis. Even in a known-cover scenario, the method has approximately 10 69238 combinations, making it practically impossible to break. REYHealth maintains moderate linearithmic time complexity ( O ( n log n )), ensuring practical efficiency without high computational cost. Comparative evaluations show REYHealth achieves the highest PSNR and SSIM, reaching 74.344 dB and SSIM closest to 1.000. These results demonstrate REYHealth’s effectiveness, robustness, and scalability for secure EHR sharing.