A Reversible and Imperceptible Watermarking Scheme for MR Images Authentication

This paper proposes a novel reversible and imperceptible watermarking method to detect intentional and accidental alterations within brain Magnetic Resonance (MR) images. Authentication data is encoded into the informative part of the medical images using reversible watermarking based on the Difference Expansion (DE) method. Experimental results demonstrate that the proposed technique, whilst completely reversible, can also produce a watermarked image with low distortion. This is realized by encoding the watermark data into 'smooth' blocks inside the informative region of the image and through eliminating the need for location map required for extracting the concealed data and reconstructing the original unmodified image. Our proposed scheme delivers highly imperceptible watermarked images, at 92.9 - 97.98 dB Peak Signal to Noise Ratio (PSNR), and exceeds the others approaches presented in previous studies. Reliability of medical images is also guaranteed through detecting subsequent modifications applied to the watermarked images. This enhanced security measure, therefore, enables the detection of image alterations, by a reversible and imperceptible approach, that may offer increased confidence in the digital medical practices.