Assessing Errors Inherent in OCT-Derived Macular Thickness Maps

SD-OCT has become an essential tool for evaluating macular pathology; however several aspects of data collection and analysis affect the accuracy of retinal thickness measurements. Here we evaluated sampling density, scan centering, and axial length compensation as factors affecting the accuracy of macular thickness maps. Forty-three patients with various retinal pathologies and 113 normal subjects were imaged using Cirrus HD-OCT. Reduced B-scan density was associated with increased interpolation error in ETDRS macular thickness plots. Correcting for individual differences in axial length revealed modest errors in retinal thickness maps, while more pronounced errors were observed when the ETDRS plot was not positioned at the center of the fovea (which can occur as a result of errant fixation). Cumulative error can exceed hundreds of microns, even under "ideal observer" conditions. This preventable error is particularly relevant when attempting to compare macular thickness maps to normative databases or measuring the area or volume of retinal features.