Software Can Evaluate Endothelium of Donor Corneas

New Vision clinic, main center for refractive surgery in France, with cutting-edge technology for all eye laser operations. Pre-op diagnosis carried out by an optometrist who specializes in measuring the visual system, refraction). Analyzing the cornea with a slit lamp. (Photo by: BSIP/Universal Images Group via Getty Images)
Researchers mechanically injured donor corneas to create a pattern of cell death, which they measured using trypan and specular microscopy to compare accuracy.

Assessing endothelial cell loss (ECL) in donor cornea tissue must be accurate to avoid graft failure in corneal transplantation. Trypan-assisted automated measurements of area of cell loss (ACL) correlate with the ACL measurements produced using specular imaging, researchers report in a study published in Cornea.

The researchers developed a custom macro to analyze ACL based on trypan blue staining to address the challenge of quantifying the amount of staining and make it practical for routine use. The macro provides fully automated segmentation, providing instant measurements of ACL. In the study, they compared the accuracy of the macro with specular microscopy, the current standard of care, to determine validity.

They evaluated 11 donor corneas with no central ECL on specular microscopy. The researchers used a cannula and Fogla deep anterior lamellar keratoplasty dissector to injure the corneas in a uniform circle. They stained them with trypan blue and imaged them with both a retroillumination lighting system and specular microscopy with wide-field imaging.

The investigators manually (man-TB) and automatically (auto-TB) measured ACL from the trypan images. Software provided automated global thresholding, segmentation and calculation of ACL. They measured and calculated ACL on specular microscopy using software to manually trace dead cells’ borders and the cell nuclei surrounding the bullseye injury.

ACL on auto-TB had a strong linear correlation with ACL on specular microscope imaging (R2=0.99, residual standard error (SE)=0.0044, P <.01). ACL on specular microscope imaging was about 21% greater based on the slope of the regression.

Specular microscopy measurements of ACL based on tracing cell nuclei (0.45±0.016 mm2) instead of cell borders were not significantly different compared with auto-TB measurement (P =.95, paired t test). Man-TB measurements were less accurate.

The researchers recommended using a correction factor in comparing ACL in trypan blue imagery and specular microscopy imagery. To produce a correction factor for stellate or convoluted injuries, they suggested pushing “out the circumference around an area of injury by a distance equal to the average radii of a single cell” so the software can predict the location of dead cell borders based on the location of trypan-stained dead cell nuclei.

“Further evaluation using endothelial injuries with different sizes and shapes may be helpful for future refinement of any correction factors,” according to the research. “Ultimately, however, there may be little need for such an approach when analyzing larger areas of endothelial injury across an entire donor cornea. Under such conditions, where the scale of ACL far exceeds the size of an individual cell, the total error will likely become insignificant.”


Bedard P, Justin JJ, Hou JH. Trypan-assisted automated endothelial cell loss measurements compared with specular microscopy. Cornea. 2021;40:1031-1035. doi:10.1097/ICO.0000000000002756