Purchase this article with an account.
Nicholas G Strouthidis, Sanchalika Acharyya, Tin A Tun, Rahat Husain, Benjamin A Haaland, Wei Xin, Jean M Mari, Shamira Perera, Tin Aung, Michael J A Girard; Visibility of the Lamina Cribrosa Using OCT: A Comparison of Devices and Techniques. Invest. Ophthalmol. Vis. Sci. 2014;55(13):912.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To compare lamina cribrosa (LC) visibility and interobserver agreement, in images acquired from glaucoma and normal subjects using 3 OCT devices, with and without enhanced depth imaging (EDI) and with adaptive compensation (AC).
60 glaucoma and 60 control subjects were imaged using Spectralis and Cirrus OCT devices both with and without EDI, and using the DR1 OCT. A horizontal B-scan was acquired through the disc center (40X scan averaging for Spectralis/Cirrus, 32X for DR1). AC was applied to reduce light attenuation and improve LC visibility (Mari, et al. IOVS 2013). A subjective grading system was designed to assess visibility of anterior and posterior LC surfaces, and LC/sclera insertions. Two expert observers (NGS, MJAG) graded the 1200 images in a randomized sequence, masked to diagnosis, device or technique. LC visibility was confined to a region limited by Bruch’s Membrane Opening. Anterior LC was graded 0 (not visible), 1 (< 25% visible), 2 (25-50%), 3 (50-75%) and 4 (>75%), posterior LC either 0 (not visible) or 1 (any region visible), and LC insertions 0 (none visible), 1 (1 insertion visible), or 2 (both visible). The effect of EDI or AC was assessed using generalized estimating equations. Visibility grades were compared using pairwise Wald tests. Agreement between observers was assessed using weighted kappa.
AC improved anterior LC visibility for any device independent of EDI. DR1 + AC and Cirrus + EDI + AC generated the highest mean grades (both 3.04), significantly better than other methods (p < 0.05). DR1 + AC was the best performing method (mean 1.12) for LC insertion visibility. Posterior LC visibility was poor regardless of method. For all LC structures, interobserver agreement was fair-moderate (kappa score range 0.45 - 0.5). No variable was associated with increased anterior LC visibility, anterior chamber depth increased LC insertion visibility for Spectralis + EDI + AC and IOP was associated with expert agreement for anterior LC for Spectralis + EDI + AC (all p < 0.05).
Applying AC, independent of EDI, outperforms standard methods for anterior LC or LC insertion visibility. Posterior LC visibility remains poor suggesting that LC thickness measurements are unreliable at present. Fair to moderate interobserver agreement suggests that caution is required when interpreting LC images, although application of AC can improve LC detection.
This PDF is available to Subscribers Only