Embedding in medical images: an efficient scheme for authentication and tamper localization

摘要

In e-healthcare applications integrity of the received information is of prime importance for ensuring the accurate diagnosis. The integrity of electronic medical record (EMR) is possible only when the medical images and other relevant data received are tamper free. Reversibility of medical images after certain degree of processing is always a desired property in medical images as it aids proper diagnosis. This paper proposes a novel reversible image authentication (RIA) scheme for tamper detection and authentication of medical images. In the proposed RIA scheme, the medical image is divided into 4×4 blocks followed by embedding fragile watermark bit (for authentication) in each of these blocks. Along with authentication, the localization of tamper is achieved accurately by using LSB embedding along with mean modification approach. The security of the watermark has been enhanced with Arnold cat map, Gray coding and AES-128 encryption. The experimental results show that the scheme offers better fragility against all intentional/unintentional signal processing attacks. The comparison results between proposed RIA scheme and similar state-of-the-art schemes show superiority of our scheme in terms of imperceptivity, fragility and tamper localization. For a payload of 16 kb, the average PSNR achieved for 50 Gray scale images of size 512 × 512 is more than 47 dB. In addition, the scheme offers very less complexity; the embedding and extraction times are around 0.6 s and 0.5 s respectively. Given the features of proposed scheme it could serve as a potential candidate for transfer of EMR in an e-healthcare system.