High-capacity multiple regions of interest watermarking for medical images

•The new scheme allows multiple ROIs of a medical image to be selected and kept lossless, while regions of non-interest (RONI) are collated for embedding watermarks.•The proposed technique has achieved high-capacity watermarking and high PSNR (peak signal-to-noise ratio).•For the embedding loop times t from 1 to 5, PSNR is improved from 12.6% to 27.7% and BPP is improved from 63% to 600%. Watermarking techniques have been widely used for copyright protection, broadcast monitoring, and data authentication in medical images to protect patient's private information against tampering by unauthorized persons. A region of interest (ROI) in a medical image contains valuable diagnostic information, and thus must be preserved to avoid medical misdiagnosis. Current state-of-the-art medical image watermarking techniques only allow single-ROI data preservation. Aiming for high embedding capacity and high image fidelity, we present a novel medical image watermarking scheme which allows multiple ROIs to be selected and kept lossless, whereas regions of non-interest (RONIs) are collated for watermark embedding. The watermark is compressed, and the embedding is conducted in frequency domain to avoid piracy. Experimental results show that the proposed technique has significantly achieved high-capacity and high-PSNR (peak signal-to-noise ratio) as compared against eight existing similar techniques. In addition, our technique allows flexible adjustment on a variable that controls the tradeoff between image fidelity and embedding capacity.