Purchase this article with an account.
Kyungmoo Lee, Hrvoje Bogunovic, Young H Kwon, Mona K Garvin, Andreas Wahle, Milan Sonka, Michael David Abramoff; Automated Identification of Retinal Nerve Fiber Bundle Connectivity in 9-field SD-OCT. Invest. Ophthalmol. Vis. Sci. 2016;57(12):5941.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To propose an automated method that can identify patient-specific retinal nerve fiber bundle connectivity in a 9-field SD-OCT scan for analysis of glaucomatous damage along the entire retinal ganglion cell-axonal complex (RGC-AC).
122×9 SD-OCT (Spectralis, Heidelberg Engineering, Inc., Germany) volumes to cover 60° on the retina were obtained from 122 eyes (67 OD, 55 OS) of 122 subjects (65.4±10.4 years, 39% male, 43, 39, 40 patients with early, moderate, advanced glaucoma, respectively). Each OCT volume consists of 768×61×496 voxels (9.53×8.07×1.92 mm3). After registering 9 OCT volumes to make a wide-field OCT volume, a retinal nerve fiber layer (RNFL) was automatically segmented using the graph-theoretic approach.1 The mean RNFL thickness for each grid sector was measured using a 24-2 Humphrey visual field (HVF) grid centered on the fovea (Fig. 1A). To detect retinal nerve fiber bundle connectivity, we used an A* graph search algorithm that is capable of finding an optimal path with the minimum aggregated cost from a starting node to one of the ending nodes. Sectors 10, 16, 17, 18, 25, 27, 34, 36, 42, 43, 44, 45 are ending nodes, and the other sectors are starting nodes. A graph was constructed by connecting each starting node (x, y) to node (x+1, y-1), node (x+1, y), and node (x+1, y+1) using arcs. The cost of each arc in the graph was calculated as follows: Cost (a, b) = (1–r2(a, b))●|RNFL(a, b)|, where r2(a, b) is the Pearson product moment correlation coefficient of the RNFL thickness between nodes a, b, and |RNFL(a, b)| is the RNFL thickness difference between nodes a, b.1. Bogunovic H, Sonka M, Kwon YH, et al. Multi-surface and multi-field co-segmentation of 3-D retinal optical coherence tomography. IEEE Trans Med Imaging. 2014;33(12):2242-53.
Figs. 1B-L show 11 typical examples with the retinal nerve fiber bundle paths of 9-field OCT volumes detected by our method, showing patient-specific nerve fiber bundle paths.
The proposed method was able to identify putative nerve fiber bundle paths in individual subjects with varying degrees of glaucoma severity from 9-field SD-OCT image analysis.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.
Figure 1. (A) 9-field OCT projection image overlaid with a 24-2 HVF grid. (B-L) 9-field OCT projection images overlaid with detected retinal nerve fiber bundle paths. The color of a line segment shows the number of overlapped paths.
This PDF is available to Subscribers Only