SECURITY ENHANCED ROI BASED REVERSIBLE WATERMARKING FOR MEDICAL IMAGES

Abstract

The ever-increasing number of digital medical images and the need to exchange them between medical experts and hospitals for a better and more precise diagnosis demand that the privacy of patients be protected. As a result, medical images sent over the internet must be watermarked with the patient's information so that doctors can validate the medical image. In this research article, a novel watermarking scheme combining two techniques, Difference Expansion and Integer-to-Integer wavelet transformation is proposed. The patient data is inserted into ROI1 using difference expansion and the ROI1 coordinates are embedded in ROI2 using Integer-to-Integer wavelet transform. The ROI2 coordinates are then embedded into predefined secret region using difference expansion technique. Simulation experiments are carried out to compute PSNR, SSIM and TPM to assess the imperceptibility of water-marked medical images. By applying various attacks on watermarked images, the robustness and reversibility of the proposed watermarking strategy is examined by computing bit error rate and normalized cross correlation co-efficient, respectively. Key words: Reversible watermarking, Medical image, Difference Expansion, Integer-to-Integer wavelet transform.