Investigating test-retest reliability of different fixation locations in directional optical coherence tomography

Abstract

Purpose : Directional optical coherence tomography (D-OCT) can be used to distinguish the outer nuclear layer (ONL) and outer plexiform layer (OPL) from the Henle’s fiber layer. In this study, we investigated the test-retest reliability of different fixation locations in delineating the ONL and OPL in D-OCT.

Methods : Thirteen healthy eyes (6 females and 7 males, aged 21-32 years) were imaged using the Heidelberg Spectralis OCT2. A standard OCT B-scan was performed to serve as a reference scan for subsequent B-scans. The B-scans were taken by translating the machine to achieve a tilted image (central fixation), and by changing the internal fixation light to either nasal or temporal sides. On each side, the participants were instructed to fixate on 3 different locations of the fixation light (Fixation 1, edge of the line closest to the shifted fixation dot; Fixation 2: center of the line; Fixation 3: edge of the line furthest to the shifted fixation dot). Each B-scan was repeated on a separate session to assess the test-retest reliability of each fixation location. All B-scans were segmented automatically on the software. Data were presented as ONL and OPL layer thicknesses (mean ± SEM) across distance from the fovea, and analyzed using linear mixed models.

Results : There were no significant test-retest differences for all fixation locations except for Fixation 2 on the temporal side (P = 0.03 for ONL; P = 0.02 for OPL). There were also no significant differences when comparing the different fixation locations.

Conclusions : Central fixation with a tilted image, Fixation 1 and Fixation 3 provide a repeatable measure of true ONL and OPL thicknesses. Although there were no significant differences between the fixation locations, from a patient’s perspective it may be most efficient to obtain a reliable D-OCT using the central fixation while the operator alters the machine’s position given it requires less subjective response. Further analysis to validate the automated segmentation for D-OCT will be warranted to improve clinical usability.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.