Repeatability of retinal nerve fiber layer reflectance intensity measurement in glaucoma

Abstract

PurposeDecreased reflectance within the retinal nerve fiber layer (RNFL), observable using optical coherence tomography (OCT), appears to be related to functional loss in glaucoma. In order to investigate this in a quantitative manner, reliable measurements of reflectance intensity are needed. However, reflectance may not only be affected by pathophysiological changes in tissue properties but also confounded by other effects of ocular media and image quality. This study assesses the repeatability of intensity measurements, before and after normalization aimed to reduce test‐retest variability.MethodsData were taken from participants with glaucoma in a test‐retest study. Each had peripapillary circle scans acquired from both eyes 5 times within 10 weeks. For each scan, the following were extracted: average RNFL thickness; RNFL intensity, defined as the mean reflectance intensity of pixels within the delineated RNFL; sub‐RNFL intensity, defined as the mean intensity of pixels between the outer boundary of the RNFL and Bruch's Membrane; and intensity ratio, defined as RNFL intensity divided by sub‐RNFL intensity. For each parameter, deviations from the per‐eye mean were calculated. The intra‐eye standard deviations (SD) were expressed as percentage of the width of the range of observed measurements.ResultsThe intra‐eye SD of RNFL intensity was 12.5% of the range. Normalization reduced the intra‐eye SD of intensity ratio to 4.9% of the range, representing a significant reduction in absolute deviations with p < 0.0001 (Wilcoxon signed rank test). RNFL thickness was more repeatable, with intra‐eye SD 0.8% of its range.ConclusionsRNFL reflectance intensity varies substantially between scans. However, dividing by the intensity of sub‐RNFL tissue greatly reduces this variability. Such normalization allows useful measurements to be obtained.