An investigation of visual field asymmetry indices for detecting early Primary Open Angle Glaucoma

Abstract

Glaucoma can be defined as group of optic neuropathies associated with structural changes to the optic nerve head and the presence of irreversible visual field loss that may derive from various pathological mechanisms. One of the commonest types of glaucoma is primary open angle glaucoma (POAG). Evaluation of the visual field is an essential component in the diagnosis and management of glaucoma. Visual field loss in POAG is often asymmetric both within and between eyes.This thesis extracts samples from a large database of visual field records collected at Manchester Royal Eye Hospital to measure the discriminatory power of various asymmetry indices in the detection of POAG.The first investigation describes and evaluates 3 new Sup/Inf hemifield asymmetry indices; hemifield mean difference, hemifield standard deviation and the number of asymmetric test pairs, for detecting POAG and compares the performance of these new indices with the well-established Glaucoma Hemifield Test (GHT). A good performance was observed for the new indices. GHT can fail to detect significant asymmetry, detected by hemifield standard deviation and number of asymmetric test pairs, when an early defect crosses the GHT sector boundaries.The second investigation looked at the performance of R/L asymmetry indices and whether these indices are superior to the previously investigated Sup/Inf hemifield indices. This study found that, the new R/L asymmetry indices did not perform better than worse eye uniocular hemifield asymmetry indices in differentiating between normal and POAG eyes. Their performance was similar to that of combined uniocular indices.The third investigation describes and measures the discriminatory power of 2 indices that quantify the clustering of Sup/Inf hemifield asymmetric test locations.The final experimental chapter determines the discriminatory power of a multi- factorial model that combines a range of asymmetry indices. In a sample selected to critically evaluate early functional loss the model correctly classified 89% of controls and 80% of POAG cases and performed better than GHT and single asymmetry analyses.

Date of Award	31 Dec 2017
Original language	English
Awarding Institution	The University of Manchester
Supervisor	David Henson (Supervisor) & Tariq Aslam (Supervisor)