IoMT encryption using homogeneous lattice of coexisting neuronal chaos

In the medical field, various medical image information of patients belongs to personal privacy, and its confidentiality guarantees face dual challenges from traditional encryption methods regarding complexity and efficiency. Due to their intricate dynamics, memristor-based Hopfield neural networks have been extensively applied in the field of securing encryption for the Internet of Medical Things (IoMT). Nevertheless, existing encryption algorithms based on memristor-based Hopfield neural networks generally suffer from issues of high computational overhead and excessive resource consumption. To address these challenges, a novel memristive hybrid synaptic dual neuron network (MHDNN) and an encryption algorithm optimized for FPGA hardware platforms are proposed. Numerical simulations show that the MHDNN exhibits heterogeneous multistability, homogeneous coexisting attractors with initial offset boosting behaviors, and diverse neuron firing patterns. In addition, the MHDNN is implemented on an FPGA digital platform, and by exploiting the properties of lattice homogeneous coexisting attractors, a low-complexity (60 Hz) block hardware encryption scheme for medical image grouping with support for dynamic key updates is designed. The performance evaluation results confirm that the dynamic key block encryption algorithm based on MHDNN not only has high-quality randomness but also significantly improves the attack resistance and encryption efficiency, thereby demonstrating excellent performance and high security in Internet of Medical Things (IoMT) image encryption applications. • A novel memristive hybrid synaptic dual neuron network (MHDNN) is proposed. • The MHDNN generates heterogeneous/homogeneous multistability, and multiple firings. • The MHDNN is implemented using an FPGA for digital circuit implementation. • A low-latency dynamic key update image block hardware encryption scheme is designed.