Metody pro detekci a kvantifikaci retinálních lézí z retinálních obrazů

Abstract

The bachelor thesis deals with the segmentation of ocular background images in children with retinopathy of prematurity because this disease is currently one of the most common visual disorders in children. The aim of this bachelor thesis is to design a complex algorithm for the segmentation of retinal lesions. The algorithm includes image preprocessing and subsequent detection of retinal lesions by the active contour method. This is followed by a statistical analysis of extracted symptoms and comparison of segmentation quality between retinal images taken on RetCam 3 and Phoenix ICON devices. Standard evaluation, geometric and intensity parameters of images taken on these devices are analyzed. From the point of view of standard evaluation parameters, Phoenix ICON appears to be a better imaging system in terms of segmentation, while in terms of intensity parameters segmentation results are better with RetCam 3 fundus camera. data technique shows better results based on the study from this bachelor thesis.