Choroidal Vessel Segmentation With Scattering Optical Coherence Angiography

Abstract

Purpose:: To demonstrate three-dimensional (3D) choroidal vessel segmentation for in vivo human retina by the scattering optical coherence angiography (S-OCA).

Methods:: A non-invasive choroidal vessel segmentation algorithm for 3D OCT volume was developed. In this algorithm, the 3D OCT intensity volume of the retina is first realigned relative to the RPE layer to avoid the effect of the signal decay through the depth. Intensity distribution at each depth plane is obtained and the threshold value of the each depth is determined to be used in the following segmentation process. A binary map is created in each depth of the OCT volume to identify the part of the choroidal vessel and superimposed with inverted rescaled intensity data to enhance the visibility of the vessels. All of the segmented en-face choroidal vessel images are combined to build 3D choroidal vasculature image. One normal subject and one patient of age related macular degeneration (AMD) were examined. 3D choroidal vasculatures were segmented from OCT volumes taken by a custom built standard SD-OCT and a custom built ultra high resolution SD-OCT. The retina of normal eye was measured with high dense scanning (1024 × 140 A-lines for 5 mm × 5 mm) and macula of an AMD patient was measured with low dense scanning (249 × 254 A-lines for the same area), the measurement time was 5.5 seconds and 2.4 seconds respectively with 27,000 A-lines/sec scanning speed.

Results:: The segemented 3D choroidal vasculature shows the volumetric characteristic. The two-dimensional (2D) projection image of the S-OCA shows similar choroidal vessels with ICGA. The S-OCA can image the slow blood flow vessels, unlike the D-OCA which is one of the OCA developed by our group. En-face slicing is possible due to the 3D structure aligned to the RPE layer and from that it is possible to resolve the tangled fine vessels appeared with blurred state in the 2D image.

Conclusions:: S-OCA enables the extraction of the choroidal vasculature from the OCT intensity data even with low dense scanning. Projection image of the 3D choroidal vasculature is comparable with FA and ICGA image. The S-OCA may be able to alternate current method ICGA as the non-invasive modality.

Keywords: image processing • choroid • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)