Repeatability, reproducibility, and agreement for three optical coherence tomography instruments

To explore the repeatability, reproducibility and agreement for three optical coherence tomography (OCT) in measuring retinal (RT) and choroidal thickness (ChT).

Sixty-eight eyes were included. The average RT and ChT were obtained with VG200I and Triton, while only the subfoveal RT (SFRT) and ChT (SFChT) were measured with RTVue. Operator 1 conducted two examinations to assess repeatability, and operator 2 conducted one for reproducibility. Built-in software was used to calculate RT-auto and ChT-auto, and custom-designed algorithms were used to calculate ChT-custom to assess interanalysis and intermachine agreements.

The intraclass correlation coefficients for the repeatability and reproducibility were larger than 0.760 and 0.940 with VG200I, 0.890 and 0.910 with Triton, and 0.960 and 0.975 with RTVue, respectively. The intermachine differences for SFRT and SFChT were - 8.82 ± 3.15 μm and - 41.89 ± 8.21 μm (VG200I vs. RTVue), -26.86 ± 3.15 μm and - 56.60 ± 7.81 μm (Triton vs. RTVue), and 18.04 ± 2.01 μm and 14.71 ± 3.02 μm (VG200I vs. Triton), respectively. The mean differences between VG200I and Triton in RT, ChT-auto and ChT-custom across the different regions ranged between 16.34 and 27.80 μm, 66.53 and 96.93 μm, and 4.48 and 18.32 μm, respectively. The mean differences between ChT-auto and ChT-custom across the different regions ranged between 29.06 and 43.71 μm for VG200I and - 48.73 and - 27.98 μm for Triton. The intermachine correlations were high, with ρ > 0.655 for all parameters.

The three OCT devices all showed great repeatability and reproducibility, while the agreements between them were poor.