Axial scaling of OCT retinal images is independent of axial length

Abstract

Purpose : Literature reports¹ of an effect of axial length (AL) on axial scaling in OCT conflict with standard models of axial scaling that are independent of optical power. We sought to investigate the effect of AL on axial scaling and identify the source of the reported effect.

Methods : A model eye was constructed comprising a 19mm lens and a translatable mount housing a glass coverslip at the retinal plane. The AL of the model eye was adjusted (~14-28mm) and the coverslip thickness was measured using the software calipers on a Heidelberg Spectralis OCT/SLO and a Bioptigen Envisu OCT. For the Spectralis, optical power-related eye data (AL, corneal curvature (CC), and refraction) were varied before acquisition of OCT scans to examine possible software sources of axial scaling error. The experiments were repeated with a Spectralis for 3 human subjects with short (AL ≤ 22.5mm), medium (22.5mm < AL ≤ 25.5mm), or long (AL > 25.5mm) eyes, and foveal thickness was measured. Using the Envisu, coverslip thickness was measured as a function of reference arm path length.

Results : Using the model eye, no effect was found between AL and coverslip thickness on the Envisu (Pearson’s correlation: p=0.30, R²=0.79, n=3) or Spectralis (p=0.32, R²=0.77, n=3; Fig. 1). Thickness measurements were unaffected by varying eye data in the Spectralis software (CC: p=0.56, R²=0.13, n=5; refraction: 1µm difference over a range of 14 diopters). The range of all measurements for 3 model eye ALs, each under 4 AL settings, was 3µm compared to the reported 22µm for a similar range of ALs.¹ Thickness measurements were unaffected by varying Envisu reference arm path length (p=0.83, R²=0.0096, n=7). ALs of the human subjects were 22.26mm (short), 24.07mm (medium), and 26.92mm (long), and foveal thickness measurements were unaffected by varying eye data (AL: n-way ANOVA, p=0.27, n=3; CC: 2-way ANOVA, p=0.23, n=2; refraction: 2-way ANOVA, p=1.00, n=2).

Conclusions : Axial scaling in OCT appears to be independent of AL. No explanation of the previously reported effect was found, but software errors were ruled out. Conclusions drawn from reports in which axial scaling was adjusted based on AL should be reevaluated.
¹Röck PMID:25298411

This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.

 

Figure 1. Discrepancy between observed model eye AL/axial scale relationship and previously reported model.