Melanin-contrast-driven 3D mapping and quantification of choroidal vasculature using polarization-diversity OCT

The choroid plays a vital role in retinal health, but its clinical imaging remains challenging due to signal attenuation and limited contrast in conventional OCT. We address these limitations by leveraging polarization-diversity OCT (PD-OCT), using melanin-specific contrast via the degree-of-polarization-uniformity (DOPU) to enable detailed segmentation and visualization of the choroid. Using a combination of multi-volume registration, vessel enhancement, and segmentation, we generated high-quality 3D vasculature maps from healthy control eyes, with the en-face projection visualized in the HSV color space, where vessel intensity and choroidal thickness were mapped to value and hue components. These served as ground truth to train a deep learning model that accurately predicts choroidal vasculature from single structural OCT volumes, enabling scalable choroidal analysis across diverse populations for broader clinical application in retinal disease management.