MediCrypt: NROI-based LSB and DCT Embedding for Medical Image Steganography
R Deeksha, Sagar A Adiga, Nisa, K Dhanush, Pallavi R Kumar · 2025
Medical images often contain sensitive patient information that needs to be securely stored and transmitted. This paper proposes a novel NROI-based steganography technique for embedding such confidential patient data into medical images without degrading their diagnostic importance. Using deep learning models, Non-Regions of Interest (NROIs), areas of the image that are not crucial for diagnosis, are identified and enabling the embedding of sensitive information within these regions. The proposed method combines traditional techniques such as Least Significant Bit (LSB) and Discrete Cosine Transform (DCT) with deep learning to embed robust data. This approach ensures that the data remain undetectable and does not interfere with the image’s clinical interpretation. The effectiveness of the method is demonstrated using a publicly available medical image dataset showcasing its ability to provide high security, imperceptibility, and robustness for data embedding. This solution offers a secure and efficient means of protecting patient data while maintaining the integrity of medical images, with potential applications in healthcare systems globally.