Adaptation of a support vector machine algorithm for segmentation and visualization of retinal structures in volumetric optical coherence tomography data sets

Abstract

Recent advances in Fd-OCT^{1, 2, 3} make possible in vivo acquisition of ultrahigh-resolution volumetric retinal OCT data in clinical settings. This technology has led to new and powerful tools that have the potential to revolutionize the monitoring and treatment of many retinal and optic nerve diseases similar to the advancement achieved in other medical areas due to the application of clinical volumetric imaging. However, in order to fully realize this potential, new tools allowing the visualization and measurement of retinal features are required. Attempts at visualization of OCT volumetric retina data have recently been presented by several groups^{4, 5, 6} and have included visualizations of highly magnified retinal structures imaged with adaptive optics (AO-OCT) systems.^{7, 8} Possible approaches to retinal layer segmentation have also been presented. ^{9, 10, 11, 12}