A high-capacity and reversible patient data hiding scheme for telemedicine. (July 2022)

• Record Type: Journal Article
• Authors: Zhang, Hua; Sun, Shihuan; Meng, Fanli
• Is Part Of: Biomedical signal processing and control. Volume 76(2022)
• Publication Date: 2022-07
• Subjects: Telemedicine -- Reversible data hiding -- Electronic medical record -- Rectangular predictor -- Patient information security; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.103706
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: Three types of rectangular predictors are designed for different kinds of interpolated pixels. Rectangular predictor based on classification method protects edges and textures of original image well. Optimal base-T selection for base-T notational system achieves a higher payload maintaining better image quality. Minimum error optimal strategy is adopted to reduce the distortion of the stego image. Abstract: The quality and efficiency of telemedicine make progress successfully due to the launch of Electronic Medical Record (EMR). However, EMR suffers information security problems such as unauthorized access, data disclosure, and tampering in telemedicine transmission. To ensure security for sensitive EMR during telemedicine transmission, a novel high-capacity and reversible data hiding scheme is proposed to conceal EMR into the medical images using rectangular predictors and optimal strategy. The clinic original image is interpolated into the cover image in which interpolated pixels are predicted by rectangular predictor to facilitate reversibility and high payload for data hiding scheme. Based on the classification idea, the proposed rectangular predictor calculates the weighted factor via local correlation to protect edges and textures reducing the appearance of common interpolation defects like blurring, jaggies, and zippers. The binary secret message is converted into a series of secret symbols in base-T notational system to balance image quality and embedding capacity, in which optimal base-T is selected adaptively by the content length of the EMR. The EMR is embedded into the cover image via finding the optimal pixel modification value in liner area for lower distortion. In addition, a fragile watermark, as a discriminator of whether medical information has been tampered during transmission, is also hidden in the cover image. Abundant experimental results demonstrate that the proposed method is superior over state-of-the-art techniques in terms of payload and image quality. High payload of 2.25 bpp for PSNR 42 dB is achieved.