Softwarové prostředí pro klasifikaci a extrakci retinálních lézí z retinálních obrazových dat s využitím prvků umělé inteligence

Abstract

The aim of this diploma thesis is to create an algorithm in MATLAB and a software environment for automated extraction of retinal lesions from retinal image data using elements of artificial intelligence and subsequent testing of the results of the proposed algorithm. The introduction deals with the basic principles and devices used for imaging the retinal area. The most important part of this work is the design of a hybrid algorithm and its subsequent objective assessment. The hybrid algorithm combines two segmentation algorithms and consists of three main parts. In the first part, an algorithm for segmentation of retinal lesions is used, which, however, in many cases also incorrectly segments areas of the vascular riverbed. For this reason, an algorithm for vascular riverbed segmentation is implemented in the second part, and both results of the segmentation algorithms are subsequently subtracted. The third part of the hybrid algorithm is postprocessing based on feature extraction. The developed algorithm was tested on a dataset provided by the Eye Clinic of the University Hospital in Ostrava. Finally, a graphical user interface in the App Designer is created.