Resource Optimized Selective Image Encryption of Medical Images Using Multiple Chaotic Systems

Medical images typically contain personal and diagnostic information about patients and are typically shared among hospitals, doctors, and patients via public networks. As a result, they must be kept secure while being stored and transported to protect the patient's privacy. However, due to the unique qualities of digital image data, such as significant redundancy and correlation between pixels and their huge size, traditional cryptographic techniques are ineffective for providing proper security while encrypting them. The necessity to develop enhanced and specialized image encryption algorithms arise as traditional techniques cease to be a trustworthy option. This problem motivated them to develop many low computational complexity methods to encrypt these medical images. In this research work, Block Cipher based Region of interest medical image encryption with multiple maps is proposed. Primarily, Region of Interest (ROI) regions are extracted with the help of a Laplacian edge detection operator, and then based on the edges in the blocks, the image is classified into wanted (ROI) and unwanted (RONI or ROB) parts. Next, the important ROI regions are permuted circularly with the help of Arnold's cat map and angle value. Then permuted ROI part is encrypted using the duffling system and unimportant regions are unaltered. Finally, the ROI encrypted image is obtained by concating the ROI encrypted part and the Region of Background (ROB) unaltered part. The encryption algorithm proposed has better security performance which is demonstrated in the experiments such as attacks on data loss, differential attacks, statistical attacks, and brute force attacks. These results show that the security in the proposed method is much better than many chaotic encryption algorithms proposed in recent times.