A hybrid medical image cryptosystem based on 4D-hyperchaotic S-boxes and logistic maps

Privacy and confidentiality are essential for any patient-related information, including medical images. In this paper, a novel image encryption technique for medical images is introduced. This method is based on a four-dimensional (4D) hyperchaotic map, used to generate four substitution-boxes (S-boxes), employed for medical image encryption. The main advantage of this new method is its sensitivity toward attacks, making it highly secure. The encryption process starts by shuffling the plain image using a three-dimensional (3D) Chen map. This step is followed by the subdivision of the image into four sub-images. The third step involves replacing the pixel values in each of the sub-images with corresponding values from one of the four S-boxes. The pre-final step is the combination of the four sub-images, followed by the diffusion of this combined image while using a one-dimensional (1D) logistic map. This results in the final encrypted image. To test the efficiency of this new encryption technique, a 256 x 256 lost block within the encrypted image are subject to different types of attacks using numerical simulation. In the simulation analysis, the encrypted images are also subjected to salt and pepper noise, with the following values: 0.005, 0.05, and 0.1. The pixel correlation coefficient for images encrypted with the tested algorithm is found to be between 0.00241 and -0.000052 in the horizontal direction, between -0.00181 and -0.000952 in the vertical direction, and between 0.00263 and -0.000071 in the diagonal direction. As for information entropy, its value is close to 8 (the ideal value), between 7.9991 and 7.9994. The Unified Average Changing Intensity (UACI) ranged between 0.2857 and 0.3938, and the Number of Pixel Change Rate (NPCR) was between 0.9958 and 0.9962. These ranges are in the proximity of the optimum values for these variables. The results of encryption of other conventional encryption techniques, such as fractional discrete cosine transform with chaotic function, image encryption in the dual domain, and hybrid chaotic DNA diffusion, were compared to those of the proposed technique, which proved to be more effective and yields better results when used for medical image encryption.