АДАПТИВНОЕ LSB-ВНЕДРЕНИЕ ДАННЫХ С ВЫСОКОЙ ЕМКОСТЬЮ В МЕДИЦИНСКИЕ ИЗОБРАЖЕНИЯ

Вопросы кибербезопасности. 2026. № 4. С. 140–145.

Гаврилов А. В., Krayushkin D. V., Никитина Н. Ю., Chepovskiy A.

The purpose of the study: the development of an approach for covert embedding of patient personal data and data integrity verification markers into medical images, ensuring no distortion of critical diagnostic information. Method: an adaptive method based on Least Significant Bit (LSB) substitution is proposed. To enhance stealth and robustness, embedding is performed selectively – not across the entire image, but in textured 8×8-pixel blocks exhibiting consistently high entropy and variance values. Block identification employs preliminary image processing with the Wiener filter, enabling selection of regions with elevated local variance (contours, organ textures) while suppressing uniform areas and additive noise. Results: the method was tested on a set of medical CT images varying in anatomy, contrast, and noise levels. Quantitative stealth assessment utilized PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural Similarity Index Measure) metrics. Container capacity was reported for each image. Scientific novelty: to develop a principle for adaptive selection of embedding zones based on the analysis of local entropy and variance, modified by Wiener filtering, for the tasks of medical image steganography.

Keywords: информационная безопасность, entropy, энтропия, целостность данных, information security, Wiener filtering, digital watermarking, data integrity, водяные знаки, medical images, LSB, медицинские изображения, LSB, фильтр Винера