A medical image crypto-compression algorithm based on neural network and PWLCM

In this work, we propose a novel medical image crypto-compression algorithm based on the Artificial Neural Network (ANN) and the chaotic system. The main objective of this algorithm is to improve the safety of medical images and to preserve the information they contain. First, ANN was used to compress the image. Then, the Arnold cat map was used to shuffle the weight matrix and Piecewice linear chaotic map (PWLCM) to modify the value of the hidden layer. The proposed algorithm was applied on medical images, of different types, such as MRI, Echographic and Radiographic images, coded on 8 bits/pixel and 12 bits/pixel. The proposed algorithm was validated over two steps. The first step was destined to show the robustness and feasibility of the encryption algorithm. Thus, several tests were carried on such as the Key space and sensitivity analysis, statistical attacks, differential analysis and the NIST statistical tests. The second step was destined to validate the compression process. In this case two main types of evaluation were applied. The first one was applied to control the robustness, through the study of the influence of the size of the sub-blocks to compress. The second was used to evaluate the quality of uncompressed images through PSNR, UIQ, SNR and Correlation factor. Experimental results, confirm the performance and the efficiency of the proposed algorithm in terms of the security and the quality of the uncompressed images.