Neural-assisted image-dependent encryption scheme for medical image cloud storage

Current medical technology evolves massive reports such as electronic patient records and scanned medical images; such reports are needed to be stored securely for future references. Existing storage systems are not feasible for massive data storage. Fortunately, cloud storage services meet the demand through their properties such as scalability and availability. Cloud computing is encouraged by amazing web innovation and modern electronic contraptions. Medical images can be stored in the cloud area, but most of the cloud service providers keep the client data in the plain text format. Cloud users need to take the responsibility to preserve the medical data with their strategy. Most of the existing image encryption solutions are vulnerable to the chosen-plaintext attack because the increasing power of computers and ingenuity of hackers are opening up more and more cracks in this mathematical armour. This paper proposes Hopfield neural network (HNN)-influenced image encryption technique to withstand against various attacks which optimize and improvise system through continuous learning and updating. These methods provide a critical security feature that adapts itself for day-to-day miracles of the real world. In this scheme, the back propagation neural network has been employed to generate image-specific keys that increase the resiliency against hackers. The generated keys are used as an initial seed for confusion and diffusion sequence generation through HNN.