A robust zero-watermarking scheme for tele-ophthalmological applications

Abstract

Medical images are watermarked with patient details to impose patient identification in tele-medicine applications. Fundus images comprise of fine features like blood vessels, macula and optic disks. Conventional watermarking methods may tamper crucial information in this case. Even minutest change can distort fine features and affect final decision. ROI (Region of Interest) selection approach also fails here as fine features of fundus images are evenly distributed and it’s almost impossible to properly define Region of Interest (ROI) and Region of Non-Interest (RONI) for medical practitioner. Hence arises an imperative necessity of a watermarking system which does not tamper the fundus images and results in no loss of medical information. This paper proposes a zero-watermarking system which uses singular value coefficients in Wavelet domain for generation of unique identification code from fundus images which is strategically combined with patient ID to generate watermark. Proposed technique suits to the specific requirements of fundus images and optimizes the conflicting requirements of traditional watermarking. The technique aims at unique identification, authentication and integrity verification of medical images. Experimental results reveal the robustness of the system against some common image processing attacks and indicate its suitability for secure exchange of medical images for tele-ophthalmological applications and for storage in large distributed medical databases.