Developing algorithms for the analysis of retinal Optical Coherence Tomography images

Abstract

Vision loss, with a prevalence loss greater than 42 million in the United States is one of the major challenges of today's health-care industry and medical science. Early detection of different retinal-related diseases will dramatically reduce the risk of vision loss. Optical Coherence Tomography (OCT) is a relatively new imaging technique which is of great importance in the identification of ocular and especially retinal diseases. Thus, the efficient analysis of OCT images provides several advantages. In this thesis, we propose a series of image processing and machine learning techniques for the automated analysis of OCT images. The proposed methodology in chapter 2 localizes different retinal layers using a modified version of active contour models. In chapter 3, we propose a method which classifies OCT images based on different pathological conditions using novel methods, e.g., transfer learning and new texture detection techniques. The proposed methods along with the clinically meaningful extracted characteristics provide numbers of applications and benefits, e.g., saving a considerable amount of time and providing more-efficient and -accurate indices for the diagnosis and treatment of different ocular diseases to ophthalmologists and finally reducing the overall risk of vision loss.