Impact of retinal pigment epithelium pathology on spectral-domain optical coherence tomography-derived macular thickness and volume metrics and their intersession repeatability.

Abstract

To determine the impact of retinal pigment epithelium (RPE) pathology on intersession repeatability of retinal thickness and volume metrics derived from Spectralis spectral-domain optical coherence tomography (Heidelberg Engineering, Heidelberg, Germany). Prospective cross-sectional single centre study. A total of 56 eyes of 56 subjects were divided into three groups: (i) normal RPE band (25 eyes); (ii) RPE elevation: macular soft drusen (13 eyes); and (iii) RPE attenuation: geographic atrophy or inherited retinal diseases (18 eyes). Each subject underwent three consecutive follow-up macular raster scans (61 B-scans at 119μm separation) at 1-month intervals. Retinal thicknesses and volumes for each zone of the macular subfields before and after manual correction of segmentation error. Coefficients of repeatability (CR) were calculated. Mean (range) age was 57 (21-88) years. Mean central subfield thickness (CST) and total macular volume were 264 and 258μm (P=0.62), and 8.0 and 7.8mm3 (P=0.31), before and after manual correction. Intersession CR (95% confidence interval) for CST and total macular volume were reduced from 40 (38-41) to 8.3 (8.1-8.5) and 0.62 to 0.16mm3 after manual correction of segmentation lines. CR for CST were 7.4, 23.5 and 66.7μm before and 7.0, 10.9 and 7.6μm after manual correction in groups i, ii and iii. Segmentation error in eyes with RPE disease has a significant impact on intersession repeatability of Spectralis spectral-domain optical coherence tomography macular thickness and volume metrics. Careful examination of each B-scan and manual adjustment can enhance the utility of quantitative measurement. Improved automated segmentation algorithms are needed.