Abstract
Purpose: :
Changes in the choroid, in particular its thickness, are believed to be of importance in the pathophysiology of a number of retinal diseases. The purpose of this study is to develop an automated choroidal segmentation approach in spectral-domain optical coherence tomography (SD-OCT) volume scans and compare its performance to manual delineation.
Methods: :
18 macular SD-OCT (1024 × 37 × 496 voxels, Heidelberg Spectralis) volumes from 18 normal subjects were obtained at the Doheny Eye Institute. A 3D graph-based multi-stage segmentation approach was used to identify the choroid, defined as the layer between the outer border of the retinal pigment epithelium (RPE) band and the choroid-sclera junction. The position of the choroidal borders and resultant choroidal thickness were compared with consensus manual delineation performed by two reading center OCT graders. B-scans in which the full-extent of the choroid could not be defined by the graders due to poor visibility were excluded from the comparative analysis.
Conclusions: :
The algorithm-defined choroidal borders appeared to consistently bias to a higher position in the z-direction compared with manually-delineated boundaries resulting in a lower choroidal thickness. However, because the difference was consistent and predictable, the thickness measurements were highly correlated, suggesting that a simple offset or correction factor could yield reliable choroidal thickness measurements.
Keywords: image processing • retina • choroid