Purchase this article with an account.
Cindy Blachly, Stuart Gardiner, Brad Fortune, Deborah Goren, Michael Whitworth, Steven Mansberger, Shaban Demirel; Signal-to-Noise Ratios for Structural and Functional Tests in Glaucoma. Invest. Ophthalmol. Vis. Sci. 2013;54(15):1885. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The utility of standard automated perimetry (SAP) for monitoring glaucoma is limited by its high variability. Structural tests, such as Optical Coherence Tomography (OCT), may be more repeatable, since they do not rely on subjective patient responses. However, it is hard to compare techniques in a fair manner due to their different units and dynamic ranges. This study uses a signal-to-noise analysis to compare variability of these techniques on a common scale.
Longitudinal data were used from 445 eyes of 227 subjects with non-endstage glaucoma enrolled in the ongoing Portland Progression Project. Subjects were tested every six months with SAP (HFA) and OCT (Spectralis) on the same day for a minimum of 5 visits. Only visits with both reliable SAP visual fields (≤15% False Positives, ≤30% False Negatives and Fixation Losses) and good quality OCT scans (≥15 Quality) were included. Retinal Nerve Fiber Layer Thickness (RNFLT) was assessed from OCT scans after manually refining the delineated layer boundaries in a masked fashion. For each eye, Mean Deviation (MD) from SAP and RNFLT TSNIT average from OCT were regressed against time, and the residuals from the trend over time were calculated. The standard deviation of these residuals was used as a measure of ‘Noise’. Three measures of ‘Signal’ were used: the range of values within the dataset (Range); the change in one year in the most rapidly changing eye based on regression of the series over time (SlopeMax); and the 10th percentile of these annual changes within the dataset (Slope10) to reduce the effect of outliers. As a secondary analysis, the process was repeated using only those sequences where the MD was never worse than -3dB, to reduce the confounding effect of variability increasing with damage in SAP.
For SAP, the noise was 0.58dB. This represented 2.5% of Range, 24.0% of the worst rate SlopeMax, and 83.2% of Slope10. For OCT, the noise was 1.70µm, representing 1.8% of Range, 22.9% of SlopeMax and 67.4% of Slope10. When only relatively healthy eyes (MD>-3) were used, the noise for SAP was 2.1% of Range, 20.5% of SlopeMax and 71.1% of Slope10, while the noise for OCT was 1.8% of Range, 22.3% of SlopeMax and 65.9% of Slope10.
OCT is slightly less variable than SAP when couched in terms of its dynamic range or the expected rate of change in glaucoma clinical patients. However, the difference between the two is small.
This PDF is available to Subscribers Only