Mapping the Nerve Fiber Layer and Ganglion Cell Layer Geometry with Spectral–Domain Optical Coherence Tomography

Abstract

Purpose: : The development of high–speed spectral–domain Optical Coherence Tomography systems allow for the introduction of new imaging modalities, including the acquisition of three–dimensional datasets. Detailed information about the retinal structure over large areas can be extracted from these 3–D datasets. The purpose of this study is to reach a quantitative and accurate understanding of the geometry of various retina layers around the optic nerve head as well as the fovea. In particular we generate and analyze thickness maps of the nerve fiber layer (NFL) and the aggregate ganglion cell plus inner plexiform layers (GCL+IPL).

Methods: : Glaucoma patients with mild to moderate visual field defects, as well as normals, were recruited for this study. Three–dimensional OCT datasets of both the area centered at the optic nerve and the fovea were produced by a prototype spectral–domain system (6–8 µm axial resolution). The typical dataset sampled a parallelepiped region of 6mm x 6mm x 2mm. 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 and the boundaries of the major anatomical layers internal to the retina.

Results: : The 3D datasets generated by our machine allows us to generate thickness maps consisting of 65536 data points distributed over a square region of the retina. This wealth of sample points creates images that can accurately describe even small features. For instance nerve fiber bundles can be followed as they radiate out from the optic nerve head, as can small bundle defects or sectors with thin bundles. The parting of the bundles at the temporal raphe is mapped out clearly. Areas with reduced ganglion cell thickness around the fovea are often evident and correlate well with NFL thinning. Particularly striking is the appearance in several eyes of a sharp temporal nasal step defect, both in the nerve fiber and the ganglion cell layer. The geometry information we obtain correlates well with perimetry data.

Conclusions: : Images obtained with a spectral domain OCT can be segmented to generate detailed thickness maps of retina layers over large areas of the retina. These maps allow for an unprecedented visualization and quantitative evaluation of the corresponding retinal structures, powerfully describing the geometry of features like arcuate fiber bundle defects and/or localized ganglion cell layer thinning.

Keywords: imaging/image analysis: clinical • nerve fiber layer • ganglion cells