Benchmark data set for glaucoma detection with annotated cup to disc ratio

Abstract

Glaucoma is a lifetime medical condition which might results in the deprivation of visual sense from an individual permanantly if remained untreated and undiagnosed at early phase. Some structural changes are observed by ophthalmologists using state of art biomedical imaging techniques i.e. Fundscopy and optical coherence tomography in retinal layers and optic nerve head of person effected by glaucoma at early phase. These structural indicators might help in diagnosis of glaucoma at early phase. Cup to disc ratio analysis using fundoscopy and analyzing retinal layer thickness using optical coherence tomography are among some of the structural changes used in glaucoma diagnosis. There are many autonomous computer aided diagnosis systems that helps ophthalmologists in analyzing the fundus and optical coherence tomography images using state of art biomedical imaging and machine learning techniques. Computer aided diagnostic systems helps in early detection of glaucoma in the areas where doctor to patient ratio is small. However, these algorithms require some annotated datasets for their evaluation and accuracy. Lack of annotated benchmark datasets with respect to cup to disc ratio for glaucoma detection has led to unavailability of comparison and evaluation of glaucoma detection algorithm globally. Proposed research aims to provide an annotated dataset with respect to glaucoma detection. Annotations are done from multiple ophthalmologists. This dataset will enable in future to measure the accuracy of proposed algorithms based on Cup to disc ratio analysis.