Purchase this article with an account.
Martin Kraus, Lauren Branchini, Tony H. Ko, Richard A. Bilonick, Yuanmu Deng, Jay S. Duker, Joel S. Schuman, Joachim Hornegger, James G. Fujimoto; Increased OCT Nerve Fiber Layer Thickness Measurement Reproducibility With Image Registration Based Motion Correction. Invest. Ophthalmol. Vis. Sci. 2011;52(14):1335.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
OCT imaging enables quantitative morphometry of Nerve Fiber Layer thickness for diagnosis and tracking progression in glaucoma. Motion artifacts and signal quality limit the reproducibility of these measurements. We apply a new motion correction algorithm that uses 2 3D-OCT data sets acquired with orthogonal raster scan patterns. A pilot study on normal subjects was conducted to evaluate if this can improve the reproducibility of NFL-thickness measurements.
Correction was performed by acquiring two 3D-OCT data sets with orthogonal raster scan patterns. An image registration based method was implemented to estimate motion in each data set. The corrected data sets were merged to create a single motion free data set with improved SNR. n=13 healthy subjects were scanned using a commercial spectral domain OCT system (Optovue RTVue). Volumes consisting of 200 by 200 A-Scans over 6 by 6 mm centered on the ONH were imaged. A standard raster scan data set was acquired along with a motion corrected data set and the acquisition was repeated 3 times. Mean NFL-thicknesses of 3.45 mm diameter virtual circumpapillary scans of original and registered volumes were measured, using a varying number of segments. Reproducibility was assessed by calculating bias and imprecision using a structural equation model. Segmentation and ONH-centering were performed using software provided by Optovue.
The registered volumes showed correction of transverse and axial eye motion and also increased SNR and reduced speckle. Global mean RNFL thickness measurement showed estimated imprecision of 2.03 um for registered vs. 3.31 um in the original volumes. For 4 segments the mean imprecision was 2.62 um vs. 4.43 um.
The motion correction method yields reduced motion artifacts and improved SNR. RNFL measurements in normal subjects show improved reproducibility. These preliminary results suggest that the developed motion correction algorithm can play a significant role in improving reproducibility of NFL thickness measurements by reducing effects from eye motion and improving image quality.
This PDF is available to Subscribers Only