Classifiction of scan location in retinal optical coherence tomography

Spectral-domain optical coherence tomography is a new imaging tool to aid the diagnosis of various diseases of the eye. Two commonly used scan patterns focus on different areas of the retina and are used to measure different properties. We developed an efficient automated classification technique that distinguishes scans of the two types so that algorithms tuned to the specific scan type can be applied during computer-aided analysis. Our algorithm differentiates between scan types based on the presence or absence of vessels converging on the optic disc. We tested its performance on an extensive dataset containing a total of 1015 scans from both healthy and diseased subjects and achieved a sensitivity of 100% and a specificity of 99.7%.