An algorithm to enhance OCT Angiography images

Abstract

Purpose : High quality angiography images are crucial in assisting diagnosis of vasculature related retinopathy. We proposed an algorithm to enhance OCT Angiography (OCTA) images. The combination of high density scans and proposed image enhancement algorithm improves the visualization of microvasculature details for large field of view (FOV).

Methods : Two prototype high-density scan patterns consisting of 490x490 (6x6mm) and 654x654 (8x8mm) A-scans with a sample spacing of 12.3 µm were implemented.Four scan patterns were used: 1) 3x3mm FOV, sample spacing 12.3 µm, 2) 6x6mm FOV, sample spacing 17µm, 3)8x8mm FOV, sample spacing 23µm, 4)12x12mm FOV, sample spacing 24 µm. Patterns 2 and 3 were used for both regular and HD scans.
Images from normal/diseased eyes were acquired at least one of the scan patterns using CIRRUS^TM HD-OCT 5000 and AngioPlex® OCT Angiography (Zeiss, Dublin, CA). Superficial and deep capillary plexus slabs (SCP, DCP) were generated using the standard CIRRUS software. The OCTA image enhancement algorithm is based on an anisotropic diffusion method using the diffusion tensor filtering to enhance the vessel structure connectivity with a smoothing effect that adapts to the underlying vascular structure. The computation is based on an iterative process to determine each successive image in the family and continued until a sufficient degree of smoothing is obtained. Three graders reviewed the en face images before and after enhancement using the following criteria: 1=much worse; 2=worse; 3=about the same; 4=improved; 5=much improved.

Results : Images from 44 subjects with normal and diseased eyes (e.g., DR, dry AMD, BRVO, etc.) using each of the four scan patterns were processed with the algorithm. The improvements were presented on both high and low density scans (fig 1). The lower and upper 90% nonparametric confidence limit from 3 graders are being reported in table 1.

Conclusions : An OCTA vasculature enhancement algorithm is presented in this abstract. The results showed a significant improvement in vessel structure connectivity and vascular granularity reduction specifically in high-density OCTA images. The enhanced OCTA images improves the visualization of microvasculature details in larger FOV.

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

 

Figure 1 SRL (a) and DRL (b) slabs of the 8x8mm HD Angiography are shown before and after enhancement using an anisotropic diffusion filter.

 

Table 1. Lower and upper 90% nonparametric confidence limit from Cirrus 5000