# Repeatability of retinal vessel flicker responses in healthy individuals

## TL;DR

This study shows that retinal vessel responses to flicker light are repeatable in healthy individuals, with raw data providing more consistent results than proprietary software.

## Contribution

The study introduces a comparison of retinal vessel response repeatability using raw data versus proprietary software outputs.

## Key findings

- Retinal vessel parameters were comparable across three different timepoints in healthy individuals.
- Raw RVA data showed better repeatability than parameters generated by the RVA software.
- Time to reach maximum dilation and constriction was least repeatable between visits.

## Abstract

To determine the repeatability of retinal blood vessel parameters (from proprietary software and the raw vessel data) measured in vivo in healthy individuals at rest, during and post flicker light (FL) provocation using a standardised protocol.

We recorded the diameters of one retinal artery and one vein in each of 33 healthy adult volunteers at rest, during and post FL provocation on two occasions using the Retinal Vessel Analyser (RVA). Repeat visits were scheduled at three different timepoints: (1) within 30 mins on the same day, (2) within a fortnight and (3) within a month (n = 11, each). All participants underwent intraocular and systemic blood pressure assessment to ensure these were comparable between visits.

Retinal vessel parameters at rest, during and post FL provocation were comparable between all pairs of visits for all three groups. Repeatability between visits was assessed using Bland–Altman Analyses and Intraclass Correlation Coefficients (ICCs). Maximum dilation for arteries and veins and maximum constriction for arteries due to FL provocation computed from raw dilation data showed better repeatability than that generated by the RVA software. Time to reach maximum dilation in arteries and veins and maximum constriction in arteries was on average comparable but least repeatable between visits.

Retinal vessel parameters computed from the raw RVA data may be superior in quality to the output from the proprietary software because the latter uses fixed narrow time‐windows to determine the parameters. Variance within healthy controls, pathology groups and repeatability parameters alongside systemic haemodynamic parameters should be considered when utilising dynamic retinal vascular parameters as study endpoints.