Abstract
Abstract: :
Purpose:To characterise the repeatability of back scatter in corneal images obtained using an OCT and determine whether normalization relative to a constant scatterer improves repeatability. Methods: 20 subjects with normal corneas (mean age = 37 ± SD =10.2) were enrolled.On 2 occasions separated by at least a day, five central corneal OCT images (Humphrey–Zeiss OCT II) were acquired in quick succession. Immediately preceding and following these measures, using identical settings, OCT images of a plastic sheet (thickness = 1.2mm) with constant scattering properties were obtained as references. For the cornea, the average integrated scatter from the 5 OCT images was calculated using custom software and the reference was the average integrated scatter from the 10 OCT images of the plastic. We examined raw corneal scatter and then normalized the corneal scatter relative to the reference and quantified test–retest repeatability of these outcome measures using coefficients of variation (COV) and intra–class correlation coefficients (ICC). Results:The COV of (standard) corneal back scatter test–retest was 3.89 and the test–retest ICC was 0.45. The normalized back scatter COV was 2.37 and the ICC was 0.58. Conclusions:Small gains in repeatability can be made in the estimation corneal back scatter using the OCT when the data are scaled relative to a constant scatterer. These results suggest that the day–day variability in back scatter measured using the OCT are primarily related to the cornea and have less to do with the technique.
Keywords: imaging/image analysis: clinical • image processing