Purchase this article with an account.
Wei Feng, Ben K Jang, Kent Pham, Qienyuan Zhou, Tony H Ko; Repeatability of retinal thickness measurements with fast 3D motion correction technology (MCT) in optical coherence tomography in normal subjects and retinal patients. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):5965.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate repeatability of quantitative retinal thickness measurements with and without fast 3D motion correction in 30 normal subjects and 30 retinal patients.
Using Avanti SD-OCT system (Optovue, Inc.), 3 scan repeats with each collecting two pairs of orthogonal (fast x, hereinafter FX and fast y, hereinafter FY) 3D volumes on the retina were performed on each subject. The scan area was 8x8mm^2 and matrix size was 320x320x640(widthxheightxdepth), with a sample resolution of 25x25x3.2 µm3. One FX and one FY volume were fed into a modified and GPU-accelerated version of the motion correction technology (MCT) proposed by Kraus et al. . The resulting MCT-corrected was used to quantitatively assess the retinal thickness map in 9 areas in the ETDRS grid. For comparison, the same analyses were performed on the selected FX volume. Multivariate analysis of variance with a linear model taking into account of 5 factors, including repeats, patient groups, retinal areas, subjects and processing methods, was performed on the measurements to assess repeatability and agreement between MCT and non-MCT corrected images.
Motion artifacts in individual scan volumes were removed or significantly reduced in all cases with an average processing time of about 10 seconds. Figures 1 and 2 show motion correction for a normal subject and a retinal patient, respectively. It is seen that MCT was able to remove the motion artifacts in single scan volumes and achieve a much better SNR (red and green arrows). The difference in retinal thickness within ETDRS grid induced by repeats and processing methods (with or without MCT) were both insignificant (p=0.95 and p=0.88, respectively).
Our modified and GPU-accelerated motion correction technology was fast and able to remove most motion artifacts without impacting the repeatability of quantitative retinal thickness analysis.<br />  Kraus M et al. Biomedical Optics Express 2014;5(8):2591-2613
This PDF is available to Subscribers Only