Assessment of Inter-Device Agreement in Quantifying Choriocapillaris Flow Deficits in Healthy Eyes.

Abstract

Purpose: To compare the computation of the choriocapillaris flow deficit obtained from three different optical coherence tomography angiography devices in healthy subjects.

Methods: Twenty-six healthy subjects were included in this post hoc analysis of data from a prospective healthy subject optical coherence tomography angiography study. Fovea-centered 6 × 6 optical coherence tomography angiography volume scans were obtained by all three devices: Heidelberg Spectralis spectral domain optical coherence tomography angiography (Heidelberg, Germany) using a full spectrum probabilistic algorithm (512 × 512 pixels); Zeiss Cirrus 5000 AngioPlex (Zeiss Meditec, Dublin, CA) (spectral domain-optical coherence tomography angiography) using the optical microangiography algorithm (350 × 350 pixels); Zeiss PLEX Elite 9000 swept source optical coherence tomography angiography (Zeiss Meditec, Dublin, CA) using the optical microangiography algorithm (500 × 500 pixels). Choriocapillaris slabs were binarized using Image J (Bethesda, MD) and the choriocapillaris flow deficit % was measured applying a Phansalkar radius of ∼20 µm, adjusted for each device based on the size and resolution of the images.

Results: Twenty-six healthy eyes of 26 subjects with no ocular or systemic disease were included in this analysis. The mean age of the subjects was 47.60 ± 13.00 (range, 26-69) years and 46.15% were females. The choriocapillaris flow deficit % measured by the devices varied significantly from each other (p < .01). The mean choriocapillaris flow deficit % measured by the Spectralis (Heidelberg, Germany) was the highest at 57.00% (±13.21%), followed by the Angioplex at 25.70% (±9.00%), and the Plex Elite at 7.62 (±3.06%). There was a significant positive correlation between choriocapillaris flow deficit % measured by swept source-optical coherence tomography angiography and spectral domain-optical coherence tomography angiography devices (Angioplex and PLEX Elite) using the optical microangiography algorithm (r = 0.78, p < .01). Of note, this correlation was better than the correlation between two spectral domain-optical coherence tomography angiography devices using different algorithms (Spectralis, Heidelberg, Germany, full spectrum probabilistic algorithm and Angioplex optical microangiography) (r = 0.389, p = .04).

Conclusion: The correlation between swept source and spectral domain optical coherence tomography angiography devices using the same optical coherence tomography angiography algorithm is better than between spectral domain-optical coherence tomography angiography devices using different algorithms.