Hiding patient information in medical images: A robust watermarking algorithm for healthcare system

Abstract

Electronically transmission of the medical images is an essential prerequisite in a Healthcare system. During transmission, the medical images can be hacked and altered either partially or in totality, unless aptly safeguarded. Digital watermarking schemes are being utilized to protect the integrity of the medical image. Digital watermarking is a technique of embedding watermark in a medical image, to maintain confidentiality and security to the inserted watermark. The watermark used in the medical image can either be an image, text, audio, or video. Electronic Patient Record (EPR) in an image form is used as a watermark. A non-blind watermarking scheme for healthcare system using Discrete Cosine Transform (DCT) is presented in this paper. Block-based Cosine transformation is applied to the host medical image to obtain different frequency coefficients. The middle frequency band of Cosine transformation is considered for embedding watermark, as this frequency band provides an extra robust resistant to watermark. Before hiding the EPR information in the image, binary EPR information is encoded using Convolution Error Correcting Code (ECC) and decoded using Viterbi decoder, to enhance the accuracy of the detection process. To provide additional robust to the coded EPR data, M-ary modulation is further applied before embedding. The imperceptibility and robust performance evaluation of the scheme is tested on standard online Digital Imaging and Communications in Medicine (DICOM) brain image database by changing the watermark embedding factor. The proposed scheme is tested using various performance measures and is observed to be highly robust and imperceptible.