Reproducibility of Foveal Avascular Zone measurements across four different Optical Coherence Tomography Angiography systems

Abstract

Purpose : Foveal avascular zone (FAZ) has been assessed by different authors using Optical Coherence Tomography Angiography (OCTA) in healthy and pathologic eyes. The aim of our study is to investigate the reproducibility of FAZ area in healthy subjects on four different Spectral Domain OCTA and calculate inter observer agreement.

Methods : This is a cross-sectional non-interventional study. A 3mm x 3mm scan pattern of the right eye was obtained for each instrument. Subjects were scanned on AngioVue (Optovue RTVue XR Avanti, Optovue Inc.) based on a split spectrum amplitude decorrelation angiography (SSADA) algorithm; Zeiss AngioPlex (Cirrus HD-OCT 5000, Carl Zeiss Meditec Inc.) based on a so called optical micro-angiography complex (OMAG) modality; a prototype Spectralis OCT2 module (Heidelberg Spectralis, Heidelberg Engineering) based on a full spectrum amplitude decorrelation algorithm; and a prototype of AngioScan (RS-3000 Advance OCT, Nidek Co., Ltd) based on a so called complex decorrelation algorithm, processing both phase and amplitude of the signal.
Superficial vascular plexus was segmented using inbuilt semi automatic softwares present on each instrument. En face OCT slabs were segmented with an inner boundary at 3 microns beneath internal limiting membrane (ILM) and an outer boundary set at 15 microns beneath inner plexiform layer (IPL).
After exporting images, two masked graders reviewed and calculated manually FAZ area using ImageJ (ImageJ 1.49v; National Institutes of Health), a publicly available image processing software.
Interclass correlation coefficient (ICC) was calculated between graders and instruments.

Results : 15 eyes of 15 subjects (mean age 27,6 SD±4,7 years) were scanned by two different operators. All images had good quality with gradable FAZ at the superficial vascular plexus.
The mean FAZ area was 0,23 mm² SD ± 0,08 for AngioVue, 0,24 mm² SD±0,08 for OCT2 and AngioPlex and 0,25 mm² SD±0,09 for AngioScan.
Inter observer agreement was excellent for all the instruments, ICC>0,98. Agreement between 4 technologists was high ICC=0,94.

Conclusions : Our results provide a significant reproducibility of FAZ measurements in healthy subjects using four different OCTA technologists. As reported in previous publications inter observer reproducibility was high using ImageJ software. Dedicated automatic softwares would be a useful implement for quantitative analysis.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.