Filtrace šumu typu speckle v ultrazvukových obrazech

Abstract

This diploma thesis deals with suppression of structural speckle noise in ultrasound images by wavelet transformation. Speckle noise in ultrasound images is characterized by a grainy character which can be treated as noise and ways to suppress it can be found, or it may be considered as a signal characterizing the structure to which it belongs. The aim of this thesis was to test the effect of noise dynamics on the efficiency of cooperative waves on the created real clinical ultrasound data. At first, the thesis deals with a comprehensive research dealing with a brief description of ultrasonography, variant types of noise and their possible filtration. The practical part of the thesis is focused on the creation of a clinical test database and furthermore, a design describing an algorithm of the chosen method is created. Primarily, this work focuses on testing the efficiency of cooperative wavelets filtering for variant types of noise, especially for speckle noise. The essence of the work is to create a test environment for dynamic simulation of the influence of the intensity of deterministic noise. The functionality of the proposed filtering method is tested on three different created databases. Filtering quality assessment is conducted based on the selected objective parameters.