Systematic ultrastructural comparison of swept-source and full-depth spectral domain optical coherence tomography imaging of diabetic macular oedema.

Background/aims: Optical coherence tomography (OCT) is commonly used to diagnose and assess diabetic macular oedema (DME). Swept-source OCT (SS-OCT) promises improved imaging depth and more independence from media opacities. Heidelberg Spectralis full-depth imaging (FDI) combines details at different depths to one representation. The aim of this study was to determine the comparability of the imaging methods concerning DME ultrastructure.
Methods: Two graders assessed the presence of typical DME phenomena in eyes with centre-involving DME on Topcon Atlantis SS-OCT and Heidelberg Spectralis FDI spectral-domain OCT (SD-OCT) B-scans. Retinal layer segmentation was corrected and choroidal layers were manually segmented. Graders measured cyst and subretinal fluid (SRF) diameters and counted hyper-reflective foci (HRF). Findings were recorded and statistically analysed.
Results: Statistically significant systematic biases (Spectralis-Atlantis) were found for the HRF count (outside the central mm, -6.39, p=0.0338), chorioretinal thickness (central mm: -35.45 µm, p=0.00034), choroidal thickness (central mm: -60.97 µm, p=0.00004) and Sattler's layer thickness (-42.69 µm, p=0.0001). Intergrader agreement was excellent or very good for posterior vitreous detachment, vitreomacular attachment (central mm) and SRF presence in both devices. Manually delineated Sattler's layer thickness showed an intraclass correlation of 0.85 with FDI SD-OCT but 0.26 with SS-OCT (p=0.003).
Conclusion: Prominent aspects such as cysts in the outer nuclear layer and SRF can be identified with comparable confidence, while a significant systematic bias underlies chorioretinal, choroidal and Sattler's layer thickness and HRF count. Specialists should use the same device at every examination during longitudinal clinical consideration or cross-sectional evaluation of these ultrastructural biomarkers.
Competing Interests: Competing interests: None declared.
(© Author(s) (or their employer(s)) 2020. No commercial re-use. See rights and permissions. Published by BMJ.)