Detection of Retinal Diseases from Fundus Images Using Deep Learning and Adaptive Histogram Equality

Authors : Ali Emre Gök, Sakir Tasdemır

Pages : 123-135

View : 86 | Download : 76

Publication Date : 2024-04-30

Article Type : Research Paper

Abstract :Recently, various eye diseases such as cataracts, diabetic retinopathy, glaucoma, macular edema, myopia, and astigmatism have been seen frequently. Cataracts, diabetic retinopathy, and glaucoma cause blurred vision, loss of vision, and blindness in cases where they are left untreated and undiagnosed. Lack of experts and equipment, hardware problems, and erroneous decisions made by experts cause problems in the diagnosis process. Because of these reasons, computer-aided diagnosis systems that can diagnose accurately are required. Deep learning algorithms performed well in the field of health, recently. These results show that deep learning algorithms can be used in the diagnosis of eye diseases. In this study, various CNN models were used for classifying eye diseases such as cataracts, diabetic retinopathy, and glaucoma from fundus images. In the image preprocessing stage, the Contrast Limited Adaptive Histogram Equalization method was used. Experimental results demonstrate that VGG16 was the most successful model among the evaluated models in this study and the Contrast Limited Adaptive Histogram Equalization method increased the performance.
Keywords : Derin öğrenme, Diyabetik retinopati, Katarakt, Glokom, Görüntü sınıflandırma