Abstract
Purpose :
Visualizing optical coherence tomography (OCT) angiography (OCT-A) volumes with choroidal neovascularization (CNV) can be challenging. To enhance CNV visualization, an improved Bruch’s Membrane (BM) segmentation algorithm with different projection range and methodology is proposed.
Methods :
The choriocapillaris (CC) slab plays an important role in OCT-A visualizations of OCT images with CNV. Often, an OCT-A projection within a thin CC slab can be used to visualize CNV. To obtain reliable BM boundary segmentation results along the CC, an automated BM segmentation methodology is proposed. Firstly, edge detection techniques are applied to obtain candidate edge segments. Secondly, BM edge segments are determined by feature analysis, and the final BM result is obtained by 3D fitting. The obtained BM result borders the CC and can delineate that projection slab. We also have proposed to use dynamic thickness determined by the result of retinal pigment epithelium (RPE) and BM to enhance the contrast in CNV areas. Different projection techniques are also applied to further emphasize the CNV area.
Results :
The improved BM segmentation algorithm was tested on three 3mmx3mm (320 a-lines x 320 b-scans) and two 6x6 (512x512) macula OCT-A scans acquired by 1050nm wavelength swept-source OCT (DRI OCT Triton, Topcon Corp., Tokyo, Japan). Comparing to manual segmentation results, signed differences (in micron) in 3x3 and 6x6 images are 6.15 ± 8.12 and -1.44 ± 7.75, respectively. Figure 1(a) shows the mean projection within a slab consisting of the CC only. The vessel structure on the left is not clear. Dynamic projection range can help increase the contrast (Fig. 1(b)). Max projection can further highlight the vessel structure (Fig. 1(c)). Figure 1(d) shows the result of integration-based projection where the CNV area is emphasized.
Conclusions :
An automated BM segmentation routine is proposed to provide reliable BM boundary to delineate the CC. Combined with dynamic projection range, the visualization of CNV area can be further improved for clinical interpretation.
This abstract was presented at the 2019 ARVO Imaging in the Eye Conference, held in Vancouver, Canada, April 26-27, 2019.