Hiding EPR and watermark in medical images using repeated pixel differencing

This research proposes a novel high-capacity-based reversible data embedding strategy for e-healthcare applications. The proposed reversible embedding scheme for electronic patient records (EPR) replaced the conventional embedding and interpolation techniques with a simple pixel-to-block-level transformation, repeated pixel differencing (RPD) and conditional pixel permutation (CPP) strategies. A fragile watermark image combined with EPR is employed to identify tamper detection in medical stego-images. In the proposed pixel-to-block conversion, a pixel is replicated to a 2 × 2 pixel block to facilitate the reversibility of EPR data. For embedding, RPD determines and adjusts the 2 × 2 pixel differences based on EPR and watermark data. Further, to ensure high imperceptibility for the human visual system proposed scheme computes the average value of pixels that must similar to the seed pixel. To maintain a histogram invariant stego-images, a permutation of 2 × 2 pixels is employed to embed additional secret data bits while exploiting the CPP strategy. The proposed scheme is extensively evaluated on + 2000 images regarding embedding capacity, perceptual imperceptibility, and tamper detection while employing different image processing attacks. Experimental results show that the proposed scheme achieved an average of 2.24 bpp with 41.27 dB PNSR on medical images and 2.49 bpp with 40.68 PSNR with general images. The proposed strategy outperformed existing approaches in terms of high embedding capacity, acceptable visual imperceptibility, and tamper detection while ensuring the 100% reversibility of EPR data in medical images.