Fig 1.
Example of macular ganglion cell inner plexiform layer (GCIPL) thickness measurement as determined automatically by optical coherence tomography: (Top maps) macular GCIPL thickness maps.
Sectoral maps shows macular GCIPL thicknesses at superotemporal, superior, superonasal, inferonasal, inferior, and inferotemporal sectors. Sectoral thicknesses are measured in an elliptical annulus with a vertical outer radius of 2.0 mm and a horizontal radius of 2.4 mm. Deviation map shows the deviation of GCIPL measurements from age-matched healthy controls, shown as red (less 1% probability), yellow (1–5% probability), green (5–95% probability), and white (more than 95% probability). Cross-sectional scans at the level of fovea shows the segmentation of GCIPL. The outer border of the RNFL (retinal nerve fiber layer) is presented as a solid purple line and the outer border of the IPL is presented as a solid yellow line.
Fig 2.
Histogram demonstrating average number of segmentation errors for each signal strength.
Fig 3.
Composite figure demonstrating various artifacts on ganglion cell-inner plexiform layer automated segmentation: (A) Both, Inner and outer layer misidentification, note the purple does not follow the contour of the inner ganglion cell layer and the yellow line does not follow the outer inner plexiform layer; (B) Blink artifact: missing retinal image due to blink; (C) Out of register artifact: shifting of the scan superiorly with loss of information of temporal area; (D) Absence of outer segmentation line; (E) and (F) Degraded image with misidentification of layers in temporal (F) and nasal (E) regions. Of note, all signal strengths of these scans are >6.
Table 1.
Percentage and characteristics of segmentation errors identified using the ganglion cell algorithm analysisa.