Diagnosis of Diabetic Retinopathy by Using Artificial Bee Colony Algorithm

Thesis Type: Postgraduate

Institution Of The Thesis: Erciyes University, Fen Bilimleri Enstitüsü, Turkey

Approval Date: 2019

Thesis Language: Turkish

Student: Hüseyin Emre Cıkıt

Supervisor: Mehmet Bahadır Çetinkaya


In order to identify the new hemorrhagic regions on retinal images, K-means clustering which is one of the most common conventional clustering methods, swarm based  artificial bee colony (ABC) and population based differential evolution (DE) algorithms have been used. First of all, in order to make the new hemorrhagic regions more visible, these regions have been clarified by the algorithms.  Then,  pixel based area calculation of the new hemorrhagic regions have been made by the algorithms in order to provide more objective data about the disease and course of disease. Finally, the performances of the improved algorithms have been compared and the results were evaluated.

In this thesis, approaches based on heuristic algorithms have been improved in order to identify the new hemorrhagic regions on a FFA retinal image and monitor the course of the disease. By means of the data with high accuracy obtained by using the heuristic methods improved a database of the disease can be created and thus the course of the disease can be followed up in certain periods.

Diabetic retinopathy (DR) is a progressive disease caused by insufficient and irregular secretion of insulin hormone and leads to deterioration in the structure of blood vessels in the retinal layer. If DR cannot be correctly diagnosed and appropriate treatment methods are not developed, the risk of vision loss arises significantly. Fundus Fluorescein Angiography (FFA) is one of the most effective methods used in the diagnosis and follow up of the disease.