Purchase this article with an account.
Corey A Smith, Michael E West, Glen Sharpe, Lesya M Shuba, Paul E Rafuse, Marcelo T Nicolela, Balwantray C Chauhan; Asymmetry analysis of optical coherence tomography angiography images of the macula in glaucoma patients and healthy controls. Invest. Ophthalmol. Vis. Sci. 2019;60(9):5544.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Quantitative analysis of optical coherence tomography angiography (OCT-A) images requires a reproducible approach that accounts for sectoral loss. We tested the hypothesis that an asymmetry parameter rather than mean analysis of perfusion density would more reliably measure the decrease in OCT-A signal in glaucoma patients.
We analyzed OCT-A macula scans (Spectralis OCT2, Heidelberg Engineering) of 103 eyes from 37 healthy subjects and 66 glaucoma patients. Each subject had a 15°×15° scan pattern (384×768 lines) centered on the fovea. Two-dimensional projection images corresponding to the superficial vascular plexus (SVP), intermediate capillary plexus (ICP), and deep capillary plexus (DCP) were exported. Analyses were performed in MATLAB to calculate perfusion density, defined as the proportion of pixels with flow (white) to the total number of pixels in the area, from the whole image (Figure). Horizontal asymmetry was calculated by first creating an 8×8 grid. Then for each cell of one hemisphere, the mean grey value was compared to the mirror-image cell in the opposite hemisphere. An unsigned flow-asymmetry index (FAI) that included perfusion density and asymmetry (1:1 weight) was calculated. Linear regression was calculated to predict either perfusion density or FAI based on visual field mean deviation.
The median (interquartile range, IQR) age was 69 (10) years for glaucoma patients and 60 (10) years for controls (p < 0.05). Median (IQR) visual field mean deviation was -1.52 (3.71) dB for glaucoma patients and 0.76 (1.02) dB for the controls (p < 0.05). Perfusion density was 13% (0.27 vs. 0.31) lower in patients compared with controls, while FAI was 17% (0.40 vs. 0.48) lower. The mean (95% confidence interval) relative difference for each measure between 2 visits was 10.4 (8.1, 12.6)% for perfusion density and 8.8 (6.5, 11.0)% for FAI (p=0.28). The strength of the linear regression was stronger for the FAI (R2=0.34) compared to the perfusion density (R2=0.22) in SVP. However, in the ICP and DCP there was a poor association with both measures, R2≤0.09.
The flow-asymmetry index was better correlated with visual field mean deviation than perfusion density in the SVP. These findings suggest that analysis of OCT-A of the macula in glaucoma patients should not be restricted to perfusion density.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.
This PDF is available to Subscribers Only