A novel medical image encryption based on cellular automata with ROI position embedded

Medical images are critical to medical diagnosis and clinical treatment. However, the transmission of medical images over the Internet without any protection may cause malicious attacks and privacy leakage. Different from the natural image, the medical image contains a large number of black backgrounds. For privacy protection, the background of the medical image is unnecessary to be encrypted. Therefore, the encryption scheme based on the region of interest (ROI) is especially fit for medical image encryption, which paves a promising way to improve efficiency without sacrificing security. However, most of the existing ROI-based encryption schemes occupy exceptional space and time to transmit the marking information of ROI. To overcome this drawback, we present a novel ROI encryption scheme based on life-like cellular automata (life-like CA) and histogram shifting (HS). In the encryption stage, the life-like CA with balanced rules is used to encrypt ROI. Unlike most known image encryption algorithms, this model can support parallel computing. Furthermore, the marking information of ROI and electronic medical record (EMR) are embedded into the cipher image, which reduce the number of bits to be transmitted. The experiments and security analysis demonstrate that the proposed scheme can effectively protect ROI of different types and sizes of medical images, and is robust in resisting various attacks.