Premium
New Image Plugin for rapid and reproducible measurement of viable corneal endothelial cell density
Author(s) -
BERNARD A,
CAMPOLMI N,
HE Z,
PATAIA G,
GAUTHIER AS,
TRONE MC,
NAGOUMFOSSO T,
PISELLI S,
ACQUART S,
GAIN P,
THURET G
Publication year - 2013
Publication title -
acta ophthalmologica
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.534
H-Index - 87
eISSN - 1755-3768
pISSN - 1755-375X
DOI - 10.1111/j.1755-3768.2013.s041.x
Subject(s) - thresholding , pixel , computer science , plug in , artificial intelligence , wilcoxon signed rank test , computer vision , image (mathematics) , mathematics , statistics , programming language , mann–whitney u test
Purpose We previously described a laboratory test based on a triple Hoescht‐Ethidium‐CalceineAM labeling coupled with image analysis in order to determine the viability and mortality of endothelial cells on whole corneas (HEC‐triple labeling, Pipparelli IOVS 2011). We then defined the new notion of viable endothelial cell density (vECD). Nevertheless the process comprised a manual linear thresholding liable to induce inter and intra‐observer variability. Aim: to present a new ImageJ Plugin with improved image thresholding allowing rapid and reproducible vECD determination Methods The endothelium of 10 human corneas unsuitable for graft was stained with HEC and its whole area observed with either a fluorescent microscope (IX81, Olympus), with motorized stage or a fluorescent macroscope (MVX10, Olympus). TIFF images acquired for H, E and C were imported in image J. Our new homemade plugin comprised contrast improvement, regional selection of pixels with similar gray levels, simplified thresholding of areas facilitated by a user‐friendly dysplay of images, possibility of manual touch up to increase accuracy. Repeated measurements of vECD were done by 5 observers to determine inter and intra rater variability, and compared to basic image thresholding Results Inter and intra rater agreement was high. Compared to the basic thresholding, standard deviation of measurements was reduced by 66% (P=0.002, non parametric Wilcoxon ranked test) Conclusion We made this new rapid and reproducible free image analysis tool available for the community to help standardize research works needing robust EC viability measurements