Join security and block watermarking-based evolutionary algorithm and Racah moments for medical imaging

Abstract

Ensuring the security of medical images containing confidential patient health information has become crucial. This paper presents a multipurpose medical image system based on a join of security and watermarking to achieve high data protection. We propose a block-splitting technique applied to large-scale cover medical images and hidden watermarks. In each subspace, we calculate Racah moments and apply a new evolutionary algorithm to automatically select the best positions for moments to be used for watermark hiding. This selection is based on a set of chromosomes with a specific coding strategy and evolutionary operators. The entire watermarked medical image is transmitted, and the extraction process is applied on the receiving side to recover the hidden watermark. Our method aims to achieve an optimal trade-off between watermarking requirements (imperceptibility and robustness) and enhance data protection by combining the proposed method with the three security services (confidentiality, integrity, and authentication). This hybrid combination uses high-performance encryption algorithms applied to the watermark components: the patient’s fingerprint, Electronic Record Patient, and doctor’s face. Watermarking performance and security services are evaluated using different medical datasets (DICOM, Figshare Brain Tumor, Covid-19, CXR, and MuRa images) in the absence of noise and against local or global noise. A comparative analysis with reliable and recent methods in medical image watermarking shows that our method achieves high performance: imperceptibility with a maximum PSNR of 46 dB, robustness with a maximum NC of 0.98, authentication with a maximum PSNR of 36.2 dB, and integrity with a maximum similarity of 0.97.