Protection of medical images through MFDCuT and SA-BSO

DOI: https://doi.org/10.1007/s11042-024-19230-8
IF: 2.577
2024-04-27
Multimedia Tools and Applications
Abstract:In recent years, the healthcare sector has witnessed a profound impact from the increased utilization of digital images, particularly in the context of telemedicine applications. However, the safeguarding of these medical images has become imperative to ensure both their integrity and authenticity. The adoption of watermarking techniques has emerged as a pivotal strategy for enhancing security. Conventional deep learning-based watermarking models, despite their advantages, often fall short in achieving simultaneous blindness, robustness, and automated embedding and extraction. This paper aims to address these limitations by presenting novel methods for medical image watermarking that prioritize both authenticity and integrity. The proposed image watermarking process consists of two primary phases: embedding and extraction. Initially, the host image undergoes segmentation into non-overlapping blocks, with the Self Adaptive-based Beetle Swarm Optimization (SA-BSO) algorithm identifying optimal blocks based on entropy calculations. The objective is to minimize entropy, thereby enhancing image quality. The selected blocks are then transformed using the Modified Fast Discrete Curvelet Transform (MFDCuT) during the embedding phase. In this embedding process, a new key is generated by incorporating a block integrity bit through the trace of Singular Value Decomposition (SVD) along with the optimally chosen blocks. Moving to the extraction phase, the recovery of the message is contingent upon ensuring both authenticity and integrity. Verification of the key employed in the extraction process is crucial for establishing authenticity, while the block integrity bit is scrutinized to ensure the integrity of the proposed model. Experimental results affirm that the devised lossless scheme not only exhibits remarkable imperceptibility and robustness but also offers reliable authentication and integrity. The performance surpasses existing schemes, making it a formidable solution for the protection of medical images.
computer science, information systems, theory & methods,engineering, electrical & electronic, software engineering
What problem does this paper attempt to address?