Secure transmission and storage of medical images remain a critical requirement in modern healthcare, where ensuring both robust security and computational efficiency is essential. To address these challenges, we propose an Adaptive Cryptographic Fusion Model (ACFM), a novel seven-stage encryption framework specifically designed for medical imaging applications. The model integrates hybrid entropy-driven pseudo-random sequence generation, multi-channel image fusion, and plaintext-driven parameterization, ensuring that the encryption process is both dynamic and highly sensitive to image content. The pipeline further incorporates multi-path substitution, dynamic propagation-based diffusion, and index-driven scrambling, culminating in ciphertext reinforced with integrity checks via hash embedding. This architecture provides strong resistance against statistical, differential, and plaintext-based attacks, while maintaining full reversibility for reliable decryption. The proposed ACFM was evaluated using benchmark datasets including Diabetic Retinopathy Detection, Skin Cancer Segmentation, and Lung Cancer Detection, and compared with state-of-the-art methods such as VFHE-DWOA, HSFQM, DNA-ECC, and CLM. Experimental results demonstrate that ACFM achieves superior performance, with Pixel-Level Reconstruction Fidelity of 98.9% at 30 iterations and maintaining above 96.7% accuracy at 200 iterations, outperforming competing methods by 2–4%. Similarly, decryption reconstruction fidelity consistently exceeded 96%, while processing time was reduced by up to 20–30%, requiring only 5.4 s at 30 iterations. These findings establish ACFM as a robust, efficient, and scalable encryption strategy, making it well-suited for secure medical image management in real-world healthcare systems.
Jasmine et al. (Mon,) studied this question.