Reproducibility of segmentation error correction in age-related macular degeneration: Stratus versus Cirrus OCT

The accuracy of retinal thickness measurement in age-related macular degeneration by optical coherence tomography (OCT) is affected by threshold algorithm line errors. The reproducibility of error correction in Stratus and Cirrus OCT should be examined.

OCT examinations of a consecutive series of 104 patients with neovascular age-related macular degeneration included in another study were reviewed. 72 eyes exhibited failures in Stratus OCT and 32 eyes in Cirrus OCT and were included in this new study. Algorithm line failures of Stratus OCT (retinal thickness program) and Cirrus OCT (Macular Cube 512×128 program) were corrected independently twice by two ophthalmologists and two residents, respectively, using the Stratus and Cirrus OCT built-in software. Reproducibility was assessed by the interclass correlation coefficient (ICC).

The corrected values of central retinal thickness were significantly lower than the automated measured values in Stratus OCT for all examiners (p<0.001), while in Cirrus OCT the differences were not significant (p=0.06-0.09). For Stratus OCT, the ICC for central retinal thickness was 0.991 and 0.997 for the experienced ophthalmologists and 0.89 and 0.97 for the residents. For Cirrus OCT, the ICC was 1.0 and 1.0 for the experienced ophthalmologists and 0.99 and 0.95 for the residents.

The reproducibility of threshold algorithm line failure correction was good overall in Stratus and Cirrus OCT and can therefore be recommended to improve retinal thickness measurement, particularly when experienced examiners perform the corrections.