Purpose:
Fixing and sectioning lead to shrinkage and other artifacts in tissue morphometrics. Non-destructive imaging of the lamina cribrosa (LC) microstructure from unfixed, uncut human eyes can be done with second harmonic generated (SHG) microscopy. From SHG images we measured LC microstructural parameters and analyzed their relationships.
Methods:
SHG images (Zeiss 510 Meta LSM, lateral resolution 1.76 to 4.97 µm/pixel) were taken of the LC of 6 normal human eyes (54 to 82 yo) at normal IOP (10 or 15 mmHg). Maximum intensity projections of each LC were split into regions (Figure). Beam thickness (BT) and pore size (PS) diameters, as well as connective tissue density (TD) were obtained through supervised scripted image processing (FIJI), whereas number of beam insertions into the canal wall (IC) was counted manually and normalized by canal wall arc length. Parameter associations were tested by ANOVA, for all eyes (pooled) as well as per eye (individual). Statistical significance was defined as P<0.001 for pooled and P<0.05 for individual.
Results:
Pooled parameter averages± SD were: PS 54.2 ±21.6 µm, BT 47.7 ±21.3 µm, IC 27.9 ±25.5 insertions/mm, and TD 51.4 ±5.5%. Pooled analyses revealed significant positive parameter associations while individual analyses were more variable. For example, in the pooled analysis PS and BT had a positive correlation, whereas individual associations were positive in 1 eye and negative in 2 eyes (+1, -2) (Fig 1C). Similarly, PS and TD (+1, -5), PS and IC (+0, -1), BT and TD (+4, -1), BT and IC (+0, -3) and IC and TD (+1, -5) contrasted with the positive correlations from corresponding pooled analyses.
Conclusions:
In the pooled analysis larger PS was associated with increased BT and higher IC, but no significant association was found with TD. One possible reason is that thicker beams and more beam insertions into the canal wall provide the structural and nutritional support needed by larger pores. The differences in parameter associations between pooled and individual eye analyses, a manifestation of Simpson’s paradox, suggest that generalizing or particularizing conclusions based on microstructural characteristics of the LC should be done cautiously.
Keywords: lamina cribrosa • image processing • optic nerve