Diagnostic Ability of Subnormal Capillary Density Area using a Reference-Based OCT-Angiography Deviation Mapping Approach to detect eyes with Diabetic Retinopathy.

Abstract

Purpose : To evaluate and compare the diagnostic abilities of subnormal capillary density area, capillary density, and FAZ metrics to detect eyes with diabetic retinopathy (DR).

Methods : 163 subjects (40 controls, 40 diabetic without DR-noDR, 39 NPDR, and 44 PDR) were imaged using a SD-OCT system (Avanti RTVUE-XR, Optovue). One 3x3mm parafoveal OCT-A scan was obtained for each subject. Individual axial length was obtained for ocular magnification correction. An OCT-A layer from the ILM to 70-µm below the posterior boundary of the IPL was used for image processing and analysis. The FAZ margin was manually segmented on the OCT-A image. FAZ metrics (area, perimeter, and acircularity index) were then computed. Capillary density was measured using a custom MATLAB algorithm. A deviation map highlighting subnormal capillary density areas <5% (yellow) and <1% (red) of a reference database was generated (Fig. B,C,D,E). Mean±SD of subnormal capillary density areas, capillary density, and FAZ metrics were calculated for all groups. One-way ANOVA was used to assess the differences between groups. Receiver Operator Characteristics (ROC) curve was used to analyze each parameter for its ability to distinguish between eyes without DR (controls and noDR) and with DR (NPDR and PDR).

Results : The NPDR and PDR groups showed significantly greater values in all parameters when compared to the control and noDR groups (Fig. A). No significant difference was observed between the Control and noDR groups in any of the parameters. Areas under the ROC curve were 0.93, 0.91, 0.88, 0.85, 0.84, and 0.72 for subnormal capillary density area <1%, subnormal capillary density area <5%, capillary density, FAZ acircularity index, FAZ perimeter, and FAZ area respectively.

Conclusions : Diagnostic abilities of subnormal capillary density area, based on our reference-based OCT-A deviation mapping approach, appears to be significantly better than that of capillary density and FAZ metrics suggesting it’s potential clinical value for detecting eyes with DR.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.

 

Fig 1. A) Table displaying each parameter’s values for all groups. B,C,D,E) OCT-A capillary density deviation map highlighting subnormal capillary density areas <5% (yellow) and <1% (red) of an age- and eccentricity-matched reference database. Subnormal capillary density areas increase with worsening diabetic retinopathy.