Vessel labeling in combined confocal scanning laser ophthalmoscopy and optical coherence tomography images: criteria for blood vessel discrimination.

Abstract

IntroductionThe diagnostic potential of optical coherence tomography (OCT) in neurological diseases is intensively discussed. Besides the sectional view of the retina, modern OCT scanners produce a simultaneous top-view confocal scanning laser ophthalmoscopy (cSLO) image including the option to evaluate retinal vessels. A correct discrimination between arteries and veins (labeling) is vital for detecting vascular differences between healthy subjects and patients. Up to now, criteria for labeling (cSLO) images generated by OCT scanners do not exist.ObjectiveThis study reviewed labeling criteria originally developed for color fundus photography (CFP) images.MethodsThe criteria were modified to reflect the cSLO technique, followed by development of a protocol for labeling blood vessels. These criteria were based on main aspects such as central light reflex, brightness, and vessel thickness, as well as on some additional criteria such as vascular crossing patterns and the context of the vessel tree.Results and conclusionThey demonstrated excellent inter-rater agreement and validity, which seems to indicate that labeling of images might no longer require more than one rater. This algorithm extends the diagnostic possibilities offered by OCT investigations.