Assessing different combinations of B-scan repetitions to improve detection of flow in optical coherence tomography angiography

Abstract

Purpose : Retinal optical coherence tomography angiography (OCTA) is based on the comparison between successive B-scans taken in the same location. Differences between these repetitions are usually due to moving blood. This study examines the effect of using longer time intervals on the detection of flow.

Methods : High resolution retinal OCTA scans usually comprise sets of 4 repeated B-scans, which may be arranged in 3 pairs and processed and averaged to make flow B-scans. In this study, the B-scans in each of these pairs were separated by a time interval (Δt₀ ≈ 3.2 msec) equal to the time between successive B-scans. In addition to this approach, we also used different pair definitions to achieve longer time intervals.
3x3mm OCTA scans were acquired with a PLEX® Elite 9000 (ZEISS, Dublin, CA) and exported as raw amplitude and phase data. 17 eyes from 14 patients were included in the analysis. These scans were processed using a custom angiography algorithm that allowed selection of the repetitions combined to produce the flow signal.
In addition to using sequential scans to calculate flow, new flow B-scans were created by averaging results from comparing repetitions 1- 3, 2- 4, and 1- 4, with time intervals 2Δt₀, 2Δt₀ and 3Δt₀ respectively. The resulting images were compared, and vessel density was calculated in the 4 zones of the inner ETDRS ring.

Results : Figure 1 shows vasculature visualized from combining pairs of temporally adjacent B-scans and the results of redefining these pairs. There is better detection of flow in small vessels, along with some increase in noise, when using pairs having larger values of Δt. Figure 2 shows increased vessel density in the latter method for this test set.

Conclusions : Using B-scan pairs with longer time intervals results in increased sensitivity to flow and better visualization of vascular structure.

This is a 2021 ARVO Annual Meeting abstract.

 

Figure 1. Above left: 3x3mm flow superficial en face projection using adjacent B-scans to form pairs. Above right: New pair definitions from the same data result in better detection of capillaries. Below: Corresponding skeletonized vessel diagrams.

 

Figure 2. Bland-Altman plot of vessel density for 17 eyes. Positive difference (Method 2-Method 1) shows Method 2 results in greater measured vessel density.