Abstract
Objective: OCT is an essential tool for evaluating retina macular pathology, however several aspects of data collection and analysis affect the accuracy of retinal thickness measurements. Here we evaluate sampling density, axial length compensation and scan centering as major factors affecting the accuracy of ETDRS retinal thickness maps.
Methods: 115 individuals were imaged using Zeiss Cirrus HD-OCT, with a subset used to examine intrasession variability. Ten patients with various macular pathology were also imaged. Down-sampled retinal volumes were derived using custom-Matlab software to examine the affect of B scan density on ETDRS thickness maps. Axial length was measured in all subjects, and the lateral scale of ETDRS maps was corrected. Lastly, the effect of centering of the ETDRS grid on the foveal center was evaluated compared to the standard protocol of measurements based on the center of fixation.
Results: Coefficient of repeatability was found to be less than 5 microns for any given ETDRS segment. In individuals with normal vision, errors of nearly 50 microns in total ETDRS thickness were observed when not correcting for differences in axial length, and over 100 microns when not centering the scan on the anatomical fovea. Errors were significantly larger in individuals with macular pathology.
Conclusion: Our data identify the number of B scans required in a macular volume to obtain an accurate ETDRS thickness map. Our data illustrate the importance of incorporating inter-individual differences in axial length in creating accurate macular thickness maps. Finally, the ability to center the macular volume on the foveal center prior to analysis results in more accurate ETDRS maps.