Abstract
Pterygium is an eye disease occurs when an overgrowth tissue encroaches into a cornea region of the eye. It commonly affects people who live near-equatorial areas such as Malaysia and being exposed to a condition with excessive wind, ultraviolet radiation or dust. One of the conventional ways of pterygium detection is a manual screening approach by ophthalmologists. Pterygium is diagnosed after a physical examination of the eyes is conducted. The eye images are photographed to monitor the growth of the pterygium tissues. If necessary, specialized diagnostic tests may be done, particularly when the pterygium extends onto the cornea regions. For instance, a corneal topography will be used to map the surface of the cornea to detect any distortions that may arise with a larger pterygium tissues growth. To the best of our knowledge, there are limited numbers of studies that applied digital image processing (DIP) approach to early detect this ocular disease using anterior segment photographed images (ASPIs). Hence, this project proposes an algorithm to identify pterygium disease using ASPIs obtained from four different databases that are UBIRIS, MILES, RAFAEL, and QPEI. The proposed screening system consists of 4 main modules namely ASPIs data collection, cornea segmentation, feature extraction and pterygium detection modules. By calculating the ratio of radius in the cornea segmented regions and using a threshold value of 1.00, the pterygium detection results give 90.60% True Positive (TP), 77.24% True Negative (TN), 22.76% False Positive (FP) and 9.40% False Negative (FN).
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call