Enhanced visualization of choroidal vasculature map using structural OCT

Abstract

Purpose : The analysis of the choroidal layer using structural optical coherence tomography (OCT) has become an important tool for the diagnosis of ocular diseases. Through generating choroidal slabs, the vasculature in this layer can be visualized. However, artifacts due to overlying superficial retinal vasculature shadows can lead to misinterpretation of the choroidal vasculature maps. We present a method to reduce the described artifacts from choroidal layer en face images for enhanced visualization and analysis.

Methods : Our method has following steps:
1) Segment retinal pigment epithelium (RPE) and choroidal-scleral boundary.
2) Create RPE retinal layer slab from 40μm above RPE to 10μm below RPE.
3) Create choroidal layer slab from 10μm below RPE to scleral-choroidal boundary.
4) Assuming that RPE retinal slab U is partially superimposed on choroidal layer slab X resulting in a choroidal layer slab with vessel shadow artifacts I, the artifact removal can be formulated as X = I – wU with w ∈[0,1] being the fraction of U superimposed on X. w is solved by minimizing the weighted average of the normalized cross-correlation square γ² of local region U_i and X_i (8x8 pixels) with the local weight of v_i=Var(U_i)Var(X_i), formulated as min_w∑_iγ²(U_i,X_i)v_i /∑_iv_i. The explicit solution is w=∑_iCov(U_i,I_i)Var(U_i)/∑_iVar²(U_i).

Performance of the algorithm was evaluated using 96 SS-OCT volume data of 500x500 A-scans over 12x12mm acquired using PLEX® Elite 9000 (ZEISS, Dublin, CA). The volume data includes normal and diseases such as Age-related Macular Degeneration.
A grader reviewed the images before and after artifact removal, and rated the level of enhancement as being (5) much improved, (4) improved, (3) about the same, (2) worse, or (1) much worse.

Results : Figure 1 shows choroidal layer slabs with vessel shadow artifacts, the corresponding RPE retinal layer slabs, and the resulting choroidal slabs without artifacts. Our results showed 88 images were graded 4 or above. 8 remaining images were graded 3. The lower and upper 95% nonparametric confidence limit of the grading are 3 and 5 respectively.

Conclusions : We present a method to reduce vessel shadow artifacts in choroidal layer slab images. Our approach is specifically useful for visualization and diagnosis of a variety of ocular diseases.

This abstract was presented at the 2019 ARVO Imaging in the Eye Conference, held in Vancouver, Canada, April 26-27, 2019.

 

Figure 1: RPE slab (left), choroidal slab image with vessel shadow artifacts (center), and choroidal slab after vessel shadow artifacts removal (right).