Laboratorní úloha - analýza vyhlazovacích filtrů pro zpracování medicínských obrazů

Abstract

This bachelor's thesis focuses on the issue of noise reduction in medical images, specifically for CT and MRI modalities. In clinical practice, noise disrupts the visualization of anatomical structures, which degrades the diagnostic value of the images. The aim is to analyze the impact of selected smoothing filters on image quality and to identify optimal methods that effectively suppress noise while preserving fine details. The experimental part was implemented using MATLAB software, in which an algorithm was designed to test thousands of combinations of noise and filters on 50 CT and 50 MRI images from the AMOS database. The success of individual filtering applications was subsequently evaluated using objective image quality metrics and further verified using Otsu segmentation. Extensive analysis proved that there is no single ideal filter for medical data. However, the best results were achieved using bilateral and guided filters. Thus, the thesis provides experimentally supported recommendations for image preprocessing.