Author: Stewart Lake
Lake, Stewart, 2022 Optical coherence tomography shape analysis and retinal disease, Flinders University, College of Science and Engineering
Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.
Optical coherence tomography (OCT) has evolved over a period of 30 years from an interesting research tool to a ubiquitous method of imaging the eye in clinical practice, now considered essential in the management of corneal, glaucoma, neuro-ophthalmic and retinal disease. Without it, the transformation of the management of medical retinal diseases (macular degeneration and diabetes) through the use of anti-vascular endothelial growth factors would have been difficult and far more complex for both patient and physician. This thesis emerged from the observation that OCT retinal shape differs between eyes, particularly with increasing myopia, leading to the question as to whether there was useful information to be learned from the retinal contour.
Both magnetic resonance imaging and more recently the OCT have already been used to analyse the shape of the posterior segment of the eye. The objectives of this thesis are to more widely explore retinal shape measured with spectral domain OCT. OCT samples an area smaller than the retinal entirety, and combined images may alter shape information, so local retinal shape is used. While many papers have described the retinal curvature of myopic eyes, here the irregularity in retinal contour is employed as a sign, describing its distribution across both the posterior and mid-peripheral retina. This irregularity is then correlated with axial length, the primary determinant of myopia. This thesis further advances the understanding of retinal shape by examining the relationship between irregularity and posterior vitreous detachment (PVD), retinal tears and retinal detachment. Using discriminant analysis, a machine learning algorithm suitable for small-medium sized datasets, evidence is found for differences in retinal irregularity between eyes that can be used to classify retinal detachment and PVD eyes. Between eye and within eye observations over time suggest these differences do not arise from the PVD event or surgery, nor are they directly associated with chorio-retinal scarring from the use of therapeutic laser. During the course of this work, swept source OCT became available. Images collected with the swept source OCT confirm the utility of retinal irregularity features as a tool for classification.
The novel shape analysis described here requires only data easily available in the clinic, and has the potential to provide quick, high resolution, cost effective local shape information relevant to eye health. As a result of this work, it is now known that quantification of peripheral retinal irregularity with OCT is not only possible but useful. Retinal irregularity is a biomarker that relates to myopia, and may improve the ability to predict retinal detachment, which opens the possibility that vision loss may be preventable in at risk eyes.
Keywords: optical coherence tomography, retinal detachment, posterior vitreous detachment, machine learning
Subject: Ophthalmology thesis
Thesis type: Doctor of Philosophy
Completed: 2022
School: College of Science and Engineering
Supervisor: Professor Karen Reynolds