Structurally enhanced perimetric examination

Abstract

Aims/Purpose: To assess the use of structural data to guide perimetric examination.Methods: Two perimetric strategies were tested on ten healthy participants: standard ZEST with population‐based prior distributions and Structural‐ZEST (S‐ZEST), enhanced with OCT data to determine the starting parameters. Both strategies were tested twice, in random order. The Compass Automated Perimeter (CMP, iCare) was used to collect the perimetric data through the Open Perimetry Interface and bespoke Shiny user interface (UI). The UI was used to import the Spectralis SD‐OCT scans (Heidelberg Engineering, Germany) and align the OCT data with the fundus image captured by the CMP. A 10–2 grid was centred on the fovea of each participant, aligned with the fovea‐disc axis. The UI calculated the Ganglion Cell Layer (GCL) thickness corresponding to each test location, after accounting for ganglion cell displacement. The starting priors for S‐ZEST were modelled using an independent dataset of glaucoma and healthy subjects. Bland–Altman plots were used to quantify test–retest variability. Bootstrap was used to calculate confidence intervals (CIs) and p‐values to compare the widths of the 95%‐Limits of Repeatability (LoR). Correlation between GCL thickness and sensitivity (dB) and the difference in test duration was calculated using linear mixed effect models. Agreement was quantified with the mean difference between the results of the two strategies.Results: 95%‐LoRs for S‐ZEST of the two devices were similar (p = 0.829). The structure–function correlation was significant for both (p < 0.001), and very similar between S‐ZEST (R2 = 0.24) and ZEST (R2 = 0.23). There was a small but statistically significant mean difference in sensitivity between S‐ZEST and ZEST (0.56 [0.35, 0.77] dB, p = 0.034). S‐ZEST was 2.7 [1.2, 3.5] min faster than ZEST (p < 0.001).Conclusions: S‐ZEST significantly shortened testing time without reducing repeatability in healthy observers.