Optical coherence tomography imaging: Automated binarization of choroid for stromal-luminal analysis

Structural changes in the choroid, interspersed between the retina and sclera, could indicate various vision impairments. So far, choroidal thickness and volume measured from optical coherence tomography (OCT) scans of the rear part of the eye have found clinical applications. However, such gross measurements do not provide vasculature-specific information crucial in managing various chorioretinal diseases. To fill this gap, we propose an automated method of ratiometric quantification of choroidal vascular (luminal) and interstitial (stromal) regions. Specifically, our method removes idiosyncratic artefacts of OCT imaging including speckle noise, exponential dynamic range compression, and depth-dependent attenuation in a targetted manner using median filtering and exponential enhancement. The proposed method assumed significance as manual estimation of the desired ratio is infeasible in view of the inherently fine structure of many choroidal vessels. Further, our technique demonstrated vast superiority over existing protocol based on the generic ImageJ software in terms of quality. Moreover, unlike the existing protocol which is interactive, our method could be applied to high-throuput and volumetric analysis.