Reproducibility of Nerve Fiber Layer Thickness Measurements with Registration Based Motion Correction for Spectral Domain OCT in Normal and Glaucomatous Eyes

Abstract

Purpose: : OCT imaging of the retinal nerve fiber layer (RNFL) is a powerful tool for the detection and monitoring of glaucoma. Motion artifacts are significant problems for ophthalmic OCT imaging and limit the reproducibility of quantitative measurements. This study evaluates the reproducibility of RNFL thickness measurements using a registration based motion correction algorithm for spectral domain OCT in normal and glaucomatous eyes.

Methods: : Randomly selected eye from 20 healthy and 25 glaucoma patients were imaged using a commercial spectral domain OCT system (RTVue, Optovue, Inc.). One standard volume of 200 by 200 A-Scans along with two orthogonally scanned volumes of 200 by 200 A-Scans were acquired over a 6 by 6 mm region centered on the optic nerve head (ONH). A registration based motion correction algorithm was applied to the orthogonally scanned volumes to generate a merged motion-corrected volume. RNFL thickness measurements were taken along a 3.45mm diameter virtual circle centered on the ONH using software provided by Optovue. Variance components were estimated using analysis of variance (ANOVA) with random effects.

Results: : The merged motion-corrected volume was undistorted by transverse and axial eye motion and showed improved image quality with reduced speckle. The global mean RNFL thickness measurement from the registered volume had improved standard deviation (SD) and coefficient of variation (CV) compared with the standard volume. In healthy subjects, the global mean RNFL thickness measurement had 0.96um SD and 1.03% CV for motion corrected vs. 1.42um SD and 1.56% CV for standard volumes. In glaucoma patients, the global mean RNFL thickness measurement had 1.31um SD and 1.79% CV for motion corrected vs. 1.58um SD and 2.20% CV for standard volumes.

Conclusions: : These results demonstrate that registration based motion correction not only improves image signal quality but also improves measurement standard deviation from volumetric OCT data. Registration based motion correction can play an important role in improving reproducibility of RNFL thickness measurements by reducing effects from eye motion and improving image quality.

Keywords: imaging/image analysis: clinical • image processing • nerve fiber layer