Enhancing security of medical images using code-based intermittent encryption and convolutional neural network

Abstract There has recently been a rise in the demand for telemedicine systems that securely and effectively transmit medical pictures. Cognitive radio (CR) significantly uses the unutilized spectrum by using the notion of spectrum sensing. Like certain other patient records, medical imaging data has strict requirements for security and anonymity. This makes sending healthcare picture information via an exposed system difficult because of the problems identified and the risks associated with massive data spillage. This study suggests a reliable CR technology with an image encryption technique to transmit medical images securely. In the proposed approach, the convolutional neural network method has been employed for complaisant spectrum sensing, where the Fusion Center trains the network for classification tasks using historical sensing data. Due to the proper training, the system runs in a time-slotted fashion. The proposed method provides an actor-critic transfer learning technique for a secondary user to select its processing method to raise confidence level while observing energy constraints. Finally, the numerical simulation results are examined to assess the suggested approaches under various configurations related to peak signal-to-noise ratio and structural similarity index which provide 90% more efficiency than the traditional simulated techniques.