Early Stage Glaucoma Detection Using Adaptive Geometric Ellipse Method

Abstract

Retinal blood vessel segmentation is of great importance in the pathological analysis of retinal images. Glaucoma is a severe and incurable disease. Early detection of the disease is, therefore, essential. In this paper, we present the early detection of glaucoma from the altered blood vessel behavior. The method consists of 2 main phases. First, blood vessel was segmented on the retinal image. Finally, the bend area was detected using principal component analysis. The algorithm is evaluated using confusion metrics, and the performance reaches 94.18% of accuracy, 51.21% of precision, 92.12% of recall.