A dynamic approach for a lightweight and secure cipher for medical images

Abstract

Protecting the contents of medical records is of paramount importance when it comes to preserving patients’ privacy. Most existing cryptographic-based solutions rely on traditional encryption algorithms having a multi-round structure, which introduces processing latency and requires increased resources. Medical images possess special characteristics compared to other types of images. The main goal of this paper is to leverage these characteristics to design and implement an efficient and secure encryption algorithm for such images. The proposed solution defines three variants of encryption algorithms: (a) full, (b) middle-full, and (c) selective. The full approach encrypts all sub-matrices of an image, while the middle-full variant is a middle solution between the selective and full algorithms and its goal is to just hide the type of the medical image. Selective encryption identifies a set of sub-matrices of an image according to a statistical average test, known as region of interest (ROI). In the three approaches, a high security level is ensured since each image is encrypted independently of the previous and next images. Also, all primitives of the proposed cipher, such as permutation and substitution, depend on a dynamic key. Furthermore, the encryption scheme is efficient since the proposed round function is lightweight and applied for only one round. This reduces the latency and the required resources as compared to traditional cryptographic schemes. The proposed approach is flexible as it can be applied in either selective, middle-full, or full mode. Also, the size of a sub-matrix is variable and can be changed according to the available memory size. Several security and performance tests are conducted to evaluate the effectiveness of the proposed solution. The results validate the robustness of the proposed scheme against almost all considered types of attacks and show an improvement in terms of latency and resources compared to current image-encryption schemes. Also, the results confirm the robustness of the proposed algorithm in protecting the contents of medical images.