Reproducibility of retinal thickness measurements using Optical Cohernece Tomography (OCT)

Purpose. To determine the reproducibility of retinal thickness measurements in early background diabetic retinopathy (BDR) and normal eyes using OCT. Methods. Five 3.20 mm vertical cross-sectional images through the center of fixation were obtained in 16 eyes with early BDR (VA‡20/40) and 18 normal eyes. Coefficients of variation (CV) were assessed for 7 locations in each eye using the manually-assisted and automated modes of the computer software. Results. The mean CV was less than 10% for all locations using the manually-assisted method except fixation in BDR eyes (CV=12.4%). The automated method was less reproducible, with a CV>10% in 2/6 and 4/6 locations outside fixation, in normal and DR eyes, respectively. There were 63 (63/630 (10%) pairs of measurements) and 103 (105/560) (18%) instances in which the manually-assisted and computer-generated measurements differed by at least 50 μm, in normal and DR eyes, respectively. Automated measurements at fixation were the least reproducible due to poor internal limiting membrane reflectivity and to a lesser extent, hard exudates in BDR eyes (2.4%). Conclusions. OCT is capable of reproducible measurements of retinal thickness in normal and pathological states. Automated measurement of retinal thickness needs to be refined and its reproducibility reassessed.