A novel chaotic hyperjerk system for enhanced DICOM image encryption using LSTM models

This paper introduces a novel DICOM image encryption system leveraging a novel 4D Chaotic Hyperjerk System (N4DCHS) and Long Short-Term Memory (LSTM) models. The proposed N4DCHS, characterized by its chaotic dynamics, serves as the foundation for generating complex encryption keys. By integrating LSTM models, the system enhances the encryption process, adding layers of complexity and unpredictability. This integration addresses the synchronization issues and sensitivity to initial conditions typically associated with chaotic systems. Our results demonstrate that the LSTM-based N4DCHS not only improves security but also ensures efficient processing times suitable for real-world medical applications. Extensive analysis, including Lyapunov exponents and bifurcation diagrams, confirms the chaotic nature of the system. Additionally, correlation analysis shows a significant reduction in correlation between adjacent pixels in encrypted images, ensuring robust security. This work paves the way for more secure and efficient encryption methodologies in medical imaging and other fields requiring high-security standards.