Correlation of Quantitative Measurements with Diabetic Disease Severity utilizing Multiple En Face OCTA Image Averaging

Abstract

Purpose : To demonstrate the effect of correlating diabetic retinopathy (DR) disease severity and quantitative measurements of the retinal microvasculature on averaged en face images using optical coherence tomography angiography (OCTA).

Methods : Twenty-eight eyes from 19 aged-matched controls, 14 eyes from 9 diabetics without DR and 63 eyes from 37 diabetics with varying levels of DR without center-involved macular edema were imaged. Images included for analysis had signal strength >7, uniform illumination, foveal centration, and minimal artifact or saccades. Spectral domain OCTA images were acquired on each eye 5 times with the Angiography 3x3 scan pattern on CIRRUS^TM 5000 HD-OCT with AngioPlex (ZEISS, Dublin, CA). Automated segmented scans were averaged individually with registration for the superficial retinal layer (SRL) and deep retinal layer (DRL). Vessel length density (VLD), perfusion density (PD) and foveal avascular zone (FAZ) area and circularity were measured on averaged versus single OCTA images and were correlated to clinical diabetic disease severity based on the International Clinical Diabetic Retinopathy Disease Severity Scale. Pearson correlation coefficient compared the two groups; univariate and multivariate linear regression with clustering at the individual level was performed to correlate quantitative metrics to DR severity and best corrected visual acuity (BCVA).

Results : 103 eyes (65 patients) were imaged as part of this study, and 84 eyes met the inclusion criteria. Correlation to DR severity scale was best for VLD in the inner ring. In the multivariate linear regression when including all available OCTA parameters, averaging did not result in an increased R² value (0.82 and 0.78 for single and averaged groups, respectively). However, FAZ size was predictive for DR severity with averaging (coefficient 6.18, p<0.001), but not with single images (p=0.98). Similarly, FAZ size was predictive for logMAR BCVA with averaging (0.211, p=0.004), but not with single images (p=0.31).

Conclusions : Visualization of vessels in OCTA images improves with averaging, as does the correlation of FAZ parameters to DR severity and BCVA. Vessel density parameters were significant predictors of DR severity in single scans even without the additional benefit of averaging.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.

 

Figure 1. Averaged versus single images of the SRL and DRL from a PDR eye