Measuring Fidelity of Steganography Approach in Securing Clinical Data Sharing Platform using Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM)

Abstract

In the vast digital landscape, the practice of data hiding finds multifaceted applications, ranging from simple hobbyist endeavors to critical tasks like safeguarding user privacy and ensuring covert data transmission. One of the gaping vulnerabilities in many contemporary systems is the transparency with which information is stored, making it easily interpretable. Such clear visibility can be a gateway for potential leaks, false portrayals, or even be manipulated for various malevolent intents. Consequently, as a countermeasure, steganography emerges at the forefront, extensively being resourceful in the revolutionized data storage concept, the cloud technology. Unfortunately, most earlier image steganography methods could only conceal one type of file, audio, text, image within an image, rendering them monodynamic. This study focuses on the novel application of steganography towards embedding information across multiple images to facilitate security of clinical data sharing platform as opposed to traditional single-image methods. The implementation was carried out using Ruby on Rails architecture, leveraging the ChunkyPNG library. With the analyses of image texture features, adaptive payload distribution strategies were devised and compared with the established single-image steganographic techniques. Interestingly, our findings show employing strategies based on texture complexity and distortion distribution greatly enhances security, making it more resilient to modern pooled steganalysis. The exceptionally high PSNR values consistently above 90dB, coupled with SSIM values nearing 1, collectively underscore the near-identical nature of our original and stego images. This convergence of both metrics emphasizes the effectiveness of our steganographic methods, suggesting minimal distortions and high fidelity. Such compelling outcomes not only validate the methodology employed but also accentuate its potential for applications demanding subtle data concealment. In essence, the combined insights from PSNR and SSIM robustly affirm the project's success in achieving high-quality steganographic results.