Abstract
Purpose :
Multiple factors including image quality (signal strength) and projection artifact have been shown to impact quantitative measurements obtained from optical coherence tomography angiography (OCT-A) images. We have published that obtaining OCT-scans from eccentric portions of the pupil (“directional OCT”) can affect retinal thickness measurements and the brightness of certain retinal layers. In this analysis, we evaluate the impact of directional OCT on vessel density measurements from OCT-A.
Methods :
In this IRB-approved study, the Zeiss PLEX Elite 9000 Swept Source was used to obtain multiple 3x3 mm (300x300) scans in normal eyes (N = 12 eyes/ 6 subjects) following dilation. Each eye was imaged using 3 different pupil entry points: 2 pupil-center scans, 1 temporal scan, and 1 nasal scan. Passing through the eccentric portions of the pupil naturally creates a tilt in the resultant OCT B-scans (with nasal and temporal scans tilted in opposite directions, thereby creating the “directional OCT” effect). The angle of the scan tilt relative to the vertical was computed for each acquisition. En face OCT-A images were generated by the instrument software from 3 different planes/levels: superficial vessel plexus (SVP), deep capillary plexus (DCP), and choriocapillaris (CC). For the DCP and CC, images were generated with and without the projection artifact removal (PAR). After elastic registration, vessel perfusion density (VPD) and vessel length density (VLD) for en face images from different orientations were computed using Otsu's threshold in ImageJ (Figure). The difference in measurements between scan orientations was correlated against the difference in scan tilt angle.
Results :
The mean (± SD) VPD and VLD of the SVP, DCP and CC for the central, nasal, and temporal scan orientations are shown in the Table, before and after PAR (for the DCP and CC). The mean difference in B-scan tilt between the nasal and temporal scans for each eye was 11.36° ± 3.27°. There was no significant correlation between the difference in scan tilt angle and the difference in VPD or VLD for SVP, DCP, and CC, before or after PAR.
Conclusions :
Quantitative vessel density metrics from OCTA images are not affected by the position of the scanning beam within the pupil (i.e. by directional OCT). Defining factors that influence quantitative OCTA metrics is of relevance to the design of acquisition protocols in clinical trials.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.