Inter-delineator Variability And Intra-delineator Reproducibility Of 3-d Fluorescent Reconstructed Volumes Of Human Optic Nerve Head

Abstract

 

Purpose:

 

To test the inter-delineator variability and intra-delineator reproducibility of 3-D fluorescent reconstructed images of normal human donor optic nerve heads (ONH).

 

Methods:

 

The trephinated ONH and peripapillary sclera from 5 normal human donor eyes were serial sectioned using an automated modified Leica RM-2265 microtome at 1μm and imaged with a Nikon AZ-100 microscope fitted with Exfo PC120 fluorescent light source and an Apogee U16M grayscale camera with 4096x4096 pixel CCD chip that yielded concurrent co-localized fluorescent images (Downs JC et al, ISER 2010). We then reconstructed the detailed 3-D ONH autofluorescence images of these eyes with 1μm section thickness and a voxel resolution of 1x1x1μm and then delineated the ONH in accordance with the guidelines set for histomorphometric delineations (Yang et al 2007). To test the inter-delineator variability, we had 3 delineators (LX, LN and BS) delineating at least 2 eyes each. To test intra-delineator reproducibility each delineator repeated individual delineation 3 times. For each 3-D reconstruction, a least square ellipse was fit to the 80 marks defining Bruch’s membrane opening (BMO), creating a BMO reference plane. For each eye, the definition and extent of the Lamina Cribrosa (LC), the LC thickness and surface depth were measured from the BMO reference plane. Statistical analysis was done using ANOVA, and global LC thickness and depths were analyzed with Holm-Sidak method to compare the means.

 

Results:

 

Preliminary results showed no statistically significant difference in the BMO reference plane within and between the delineators (p>0.05). For all the 5 eyes, the mean LC thickness and depth measurements were not significantly different (Pearson product moment correlation, p>0.05) with the 3 repeated sessions of delineations by each delineator. Between the delineators, there was a significant difference in LC thickness in 3 of the eyes (Table 1). However, the LC depths were similar in 3 of the eyes between the delineators.

 

Conclusions:

 

We report that there was no significant variation in the mean LC thickness and depth measurements within an individual delineator, though there are some significant variations between the delineators in these eyes.  

 

Keywords: computational modeling • lamina cribrosa • optic nerve