Light‐Field imaging of the human optic nerve head at different wavelengths—a case study

Abstract

PurposeOne strong risk factor for glaucoma is a suspicious optic nerve head appearance with an abnormal cupping or an increase in the cup‐to‐disc ratio. To measure such geometric characteristics, typically scanning systems like OCT or SLO are used. We present that one‐shot 3D retinal imaging and measuring with a commercial fundus camera using an industrial light‐field camera is possible and works best in the green spectrum.MethodsTo demonstrate 3D imaging at the human optic nerve head (ONH), we examined one eye with suspected glaucoma (patient:37 years, male). ONH centered 30° images of the right eye were taken after pupil dilatation with Tropicamid with a Zeiss FF450 fundus camera (Carl Zeiss Meditec AG, Germany), to which an industrial light‐field camera (Raytrix GmbH, Germany) was attached. The images were taken at 450 nm, 520 nm, 600 nm and 650 nm centroid wavelengths, 40 nm bandwidth each. The maximum depth of the papilla was determined in virtual depth (VD) using the Raytrix software (RxLive, Raytrix GmbH, Germany). The results were compared among each other concerning the vessel contrast, quantity of measurement points and subjectively where they were found.ResultsThe light‐field images showed clearly the ONH in 3D. The highest image contrast was found at 450 nm, whereby the contrast decreases with increasing illumination wavelengths. Most measurement points around the papilla were found at 520 nm followed by 450 nm with a slightly lower number. At 600 nm, the number of measurement points was significantly reduced. Worst results were found at 650 nm. Plausible measurement points were only found in image regions with high contrast. Those regions were vessel borders and the papilla border.ConclusionsUsing the light‐field technique in fundus photography, it is possible to image and to measure the ONH in 3D. Screening for glaucoma and glaucoma diagnosis can be much improved taking into account that, in principle, classical fundus imaging offer only 2D data.