Privacy protection scheme for batch medical images based on double memristor cellular neural network

Combining the characteristics of large data volume and high continuity of homochromatic pixels in medical images, an effective privacy protection scheme for batch medical images is proposed based on a double memristor cellular neural network (DMCNN). The batch images are first preprocessed into an image cube, and then combined with the chaotic sequences generated by DMCNN for image scrambling and diffusion. In order to fully utilize the spatial advantages of the image cube, the scrambling process is divided into three steps: cross-plane random position column swapping, cross-plane random position row swapping, plane swapping and rotation. Then, the image cube is segmented by picking a random position using a chaotic sequence. Combined with the pixel distribution characteristics of medical images, the two parts after segmentation are diffused using the opposite order. The selected random location provides greater randomness to the privacy protection scheme, thus securing the medical images. Further, the feasibility, security, and real-time of the scheme are thoroughly tested and analyzed in simulation tests, security tests, and time tests, which proves that the proposed scheme can protect medical image information security in batch, and the efficient encryption and decryption also makes it suitable for real medical information privacy and confidentiality scenarios.