Extracting the Optic Disk Endpoints in Optical Coherence Tomography Data

Optical coherence tomography (OCT) uses backreflected light to provide micron-resolution, cross-sectional scans of biological tissues useful in diagnosing, monitoring, and studying glaucoma. The ratio of the area of the central portion of the optic nerve (the cup) to that of the complete nerve (the disk) is an indicator of ocular health. Measuring this cup-to-disk ratio accurately requires precise detection of the cup and disk limits in scans traversing the optic nervehead. A hybrid edge-supported boundary detector first finds the contour separating the retina and optic nervehead (on one side) from the vitreous humor on the other. We then partition this boundary using a model-based technique to identify the optic cup. The lower retinal-choroid boundaries are extracted using Markov models first derived by Koozekanani et al.. The endpoints of these lower boundaries represent the limits of the optic disk. Points along the Markov path are then represented in a feature space comprising two low order unitary transform coefficients and a measure of differential edge strength. By minimizing the trace of the within-class scatter matrices for optic nerve and retina, optimal estimates of the nervehead extent are found. We present thorough experimental results to establish the stability and robustness of the technique, as well as its performance limitations.