Privacy Protection of Medical Data Based on Multi-Scroll Memristive Hopfield Neural Network

• Published in: IEEE transactions on network science and engineering Vol. 10; no. 2; pp. 845 - 858
• Main Authors: Yu, Fei, Shen, Hui, Yu, Qiulin, Kong, Xinxin, Sharma, Pradip Kumar, Cai, Shuo
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.03.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Biological neural networks; Biomedical imaging; Computed tomography; Electromagnetic radiation; Encryption; Entropy (Information theory); Field programmable gate arrays; FPGA; Health services; Histograms; image encryption; Image transmission; Internet of Things; Internet of things (IoT); Medical imaging; memristive Hopfield neural network (MHNN); Memristors; multi-scroll; Neural networks; Neurons; Pixels; Privacy; Queueing; Security; Telemetry
• ISSN: 2327-4697; 2334-329X
• DOI: 10.1109/TNSE.2022.3223930

Memristive Hopfield neural network (MHNN) has complex dynamic behavior, which is suitable for encryption applications. In order to ensure the information security of the medical data transmitted in Internet of Things (IoT), we propose three new MHNN models by using a non-ideal flux-controlled memristor model with multi-piecewise nonlinearity. In these models, there are complex dynamical behaviors such as coexisting attractors, multi-scroll attractors and grid multi-scroll attractors. In terms of hardware, the proposed model is implemented using field programmable gate array (FPGA). In addition, we provide a complete set of medical data sharing solution, which are helpful for the referral patients to receive timely medical treatment. The whole solution is successfully verified on Raspberry Pi, the encrypted Computed Tomography (CT) image is transmitted safely under Message Queuing Telemetry Transport (MQTT) protocol, and the CT image is subjected to basic security analysis. The results show that the ciphertext histogram is evenly distributed, the correlation between adjacent pixels is almost 0, the information entropy reaches 7.9977, and the values of number of pixels change rate (NPCR) and unified average change intensity (UACI) are 99.6078% and 33.4875%. The solution not only performs the exchange of medical data, but also protects the privacy of patients.