Abstract
Abstract: :
Purpose: Recent advances in the field of optical coherence tomography (OCT), in particular the development of high–speed spectral–domain systems, allow for the introduction of new imaging modalities, including the acquisition of three–dimensional data sets. The problem of extracting clinically useful, accurate, robust, reliable, quantitative information out of these large data sets is rather complex. The purpose of this study is to demonstrate the capabilities of new image–analysis algorithms applied to suitable three–dimensional datasets acquired by a prototype spectral–domain OCT. Our system can automatically and accurately map the three–dimensional geometry of the retina and relate it to clinically available retinal landmarks. Methods: Three–dimensional OCT datasets were produced by a prototype spectral–domain system designed for clinical use. They consist of 128 equally–spaced B–scans covering a square region of 4mm x 4mm. The images were processed using a new iterative boundary detection algorithm. The algorithm is able to locate automatically and/or interactively several retinal features, including the global boundaries, the boundaries of the major anatomical layers internal to the retina, and the boundaries of cystic spaces. Results: Several patients with macular disease were imaged and three–dimensional retinal maps were obtained from their spectral–domain OCT datasets. In particular we successfully computed images of the vitreo–retinal interface, the retinal pigment epithelium (RPE), and cystic volumes in three–dimensional space. To evaluate the validity of our maps we analyzed their agreement with boundaries determined by visual inspection. In each case such agreement was determined to be good, meaning they agree on at least 99% of the pixels and never diverge by more than 10% of the correct local retinal thickness. Combining our retinal maps with a novel technique that produces fundus–like images directly from the OCT dataset we can precisely register the retinal geometry with the information derived via other retinal imaging modalities in common clinical use. Conclusions: Three–dimensional OCT images promise an unprecedented ability to visualize in vivo the morphologic and geometric changes associated with retinal disease. We have developed a system that can produce accurate images of the retinal geometry in three–dimensional space. In particular we can analyze the complex geometry/topology typical of many macular pathologies.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • retina • imaging/image analysis: clinical