Figure 1.
An example of conventional circumpapillary retinal nerve fiber layer (cpRNFL) analysis as provided by Cirrus HD-OCT.
(A) Overlay of RNFL thickness deviation map on the OCT fundus image with focal wedge defect (red arrows) predominantly outside the 3.4 mm diameter circle sampling (red circle). (B) Corresponding 2D RNFL thickness map, RNFL focal defect is marked with red arrows. (C) cpRNFL thickness profile along the 3.4 mm diameter circle is within the normal range (green range). Red arrow pointing to the approximate location of the RNFL wedge defect. (D) The RNFL thickness measurement is summarized in 4 quadrants and 12 clock hours with all sectoral measurements within the normal range.
Figure 2.
Normative database normalization with 46 healthy eyes.
(A, B) mean and standard deviation (SD) of retinal nerve fiber layer (RNFL) thickness measurement at each sampling point (A-scan), without normalization. (C, D) mean and SD of RNFL thickness measurement after normalizing individual’s retinal nerve fiber bundle path location to population’s average location. The variations of RNFL thickness were larger at superior temporal and inferior temporal regions (brighter blue in B) because of the population variation of the bundle locations. After aligning the bundle locations and normalizing the RNFL thickness map, the RNFL thickness variations at these two regions were markedly reduced (dark blue in D).
Figure 3.
Flowchart of converting a 3D OCT image into a 2D feature map.
Figure 4.
Super pixel segmentation on 3D OCT images.
(A) Analysis output in a healthy eye, (B) glaucoma suspect, and (C) glaucomatous eye. Abnormally thin retinal nerve fiber layer is marked with small super pixels.
Table 1.
Super pixel features used as inputs for the machine learning classifier.
Table 2.
Characteristics of the study participants.
Figure 5.
Normalized histogram distribution of super pixel.
(A) RNFL thickness and (B) super pixel size for healthy (H), glaucoma suspect (GS), and glaucomatous (G) eyes.
Figure 6.
The receiver operating characteristic curves (ROCs) computed with the machine classifier method and Cirrus HD-OCT software generated mean cpRNFL thickness.
(H) healthy eyes, (G) glaucomatous eyes, (GS) glaucoma suspect eyes.
Table 3.
Area under the receiver operating characteristic curves (AUCs) of conventional circumpapillary RNFL (cpRNFL) thickness measurements in global and four quadrants from Cirrus HD-OCT software.
Table 4.
Area under the receiver operating characteristic curves (AUCs) computed with machine classifier and Cirrus HD-OCT software generated mean cpRNFL thickness.
Table 5.
Area under the receiver operating characteristic curves (AUCs) computed with machine classifier and Cirrus HD-OCT software generated cpRNFL thickness in inferior quadrant.