Computer Aided Detection of age related macular degeneration in retinal images

This work aims to realize Age related macular degeration (ARMD) detection process on the retinal images obtained by the Fundus Floresein Angiography (FFA). Critic area process has been performed by Computer Aided Detection (CAD) system which was detected on data sets generated with the use of 87 images of total. The purpose of this work, regions of interest affected by ARMD disease is to provide detection with CAD system. Thus, monitoring of the treatment of the patient can be made by doctors labeled on retinal images. This study intends to provide the detection of ARMD by separating structure like blood vessels, optic disc from retinal images using pre- processing techniques as bands of color separation, histogram processes of images. In CAD pre- processing stage the areas that can be ARMD are made to be more clearer and sharpener and edge filters and dilation algorithms are used to perform successful segmentation process. At the end of the pre-processing and segmentation stages, regions of interest are labeled based on feature extracted. Regions of interest are issued to characteristics, optic disc has been eliminated by the algorithm developed. In the final stage the regions of interest are labeled according to these features. Accuracy of system is tested by ophthalmologist as controlling the ARMD and healthy retinal images labeled by CAD process. Finally, 74 (TP and TN) positive, 13 (FP and FN) negative results in detection were reached with the developed CAD system. In detection of ARMD study, using performance evaluation criteria, the accuracy of the algorithm is obtained as 85,05%.