Repeatability and inter-observer variation of choroidal thickness measurements using swept-source optical coherence
Anders Asmussen1, Bjarke S Smith1, Flemming Møller2
1University of Southern Denmark, Odense, Denmark.
Purpose:
To estimate repeatability and inter-observer variation of choroidal thickness measurements in myopic Danish children aged 7-14 years using swept-source optical coherence tomography (SS-OCT).
Methods:
Thirty-nine children were enrolled. Optical correction was single-vision spectacles (SVS) or orthokeratology lenses (OKL). Three repeated 7 × 7 mm 3-dimensional SS-OCT macula scans and three repeated 9 mm SS-OCT line scans were collected for each child using a DRI OCT Triton. Choroidal thickness was measured using three different methods: line scan method 1 (LM1), line scan method 2 (LM2) and 3D macula scan method (3DM). Segmentation was adjusted if needed. Coefficients of repeatability (CR) and limits of agreement (LoA) were calculated.
Results:
The CRs ranged from 13.4 to 23.9 µm, 14.5 to 26.2 µm and 5.2 to 10.7 µm for LM1, LM2 and 3DM, respectively. The LoA ranged from -22.9 to +31.5 µm, -23.3 to +32.2 µm and -10.2 to +12.4 µm for LM1, LM2 and 3DM, respectively. Segmentation was adjusted in most scans (63%-92%). Mean choroidal thickness ranged from 142.2 ± 47.2 to 253.8 ± 60.9 µm and 190.1 ± 64.0 to 299.0 ± 55.8 µm for the SVS and OKL groups, respectively, measured by 3DM.
Conclusion:
The 3DM was the most repeatable method in this paediatric population. It yielded a CR of 10.7 µm, confidence interval 2.4 µm, which makes the minimal detectable difference between two measurements 13.1 µm. Most inter-observer variation could be explained by the intramethod variation. Segmentation adjustment on 3D macula scans did not increase CR on a group level.
