Метод підвищення чіткості дрібних деталей біомедичних зображень із використанням графового підсилення

Abstract

Improving the quality of biomedical images is critically important for ensuring reliable visualization of anatomical structures and pathological changes, which directly affects the accuracy of diagnosis. Improving contrast, reducing noise, and enhancing informative details contribute to more reliable image analysis by both doctors and automated data processing systems. The article presents the implementation of the proposed algorithm for improving the quality of medical images, which is based on graph edge-aware enhancement of structures. The proposed approach uses the green channel as the main source of structural information to build a graph model of local connections between pixels and performs iterative smoothing, taking into account intensity differences. Based on the difference between the initial and smoothed images, a detail map is formed, which, together with an additional high-frequency component obtained using a Gaussian filter, is used to enhance the vascular pattern. The selected details are transferred to the RGB color space with different weighting factors, which ensures the preservation of the natural color balance. A series of test images from open sources are processed and the results of the algorithm are evaluated. The results obtained demonstrate an increase in the informativeness of images and improved visualization of vascular structures. The practical significance of the approach lies in the possibility of its application in computer diagnostics systems to improve the quality of medical image analysis.