<![CDATA[Medical images transmitted over public networks are vulnerable to interception because of their large volume and strong spatial correlation among neighboring pixels. Conventional ciphers such as AES are not specifically designed for this redundancy and may leave residual statistical structure when applied directly to images. This study evaluates a hybrid chaos-based image encryption scheme that combines Arnold’s Cat Map for pixel-position confusion with the Logistic Map for value diffusion. Instead of introducing an additional chaotic map, the diffusion stage is strengthened using a double-pass cipher-block chaining (CBC) mechanism with modulo addition. The method was tested on 20 grayscale brain-tumor images (256×256 pixels) from a public medical dataset. Security was assessed using histogram analysis, adjacent-pixel correlation, information entropy, differential NPCR and UACI, key sensitivity analysis, and execution time. A baseline ACM–Logistic XOR scheme was also compared under the same key parameters. The proposed CBC-enhanced scheme achieved an average encrypted-image entropy of 7.997, correlation coefficients close to zero, differential NPCR of 99.35%, and differential UACI of 32.84%. Key sensitivity tests yielded NPCR values above 99.6% for perturbations of the logistic parameters as small as 10-14, and multi-delta analysis confirmed stable avalanche behavior down to 10-15. Decryption with the correct key recovered all images without pixel loss. These results indicate that double-pass CBC can improve differential diffusion in a lightweight two-map architecture suitable for protecting grayscale medical images.]]>
