In vivo Quantification Of Corneal Lesions Sizes Using Improved Image Analysis Software

Abstract

Abstract: : Purpose: We report a new method of in–vivo quantification of mouse corneal lesions using improved image analysis software, taking into account the three–dimensional shape of the cornea Methods:Mathematical analysis were performed to calculate the ratio between the three–dimensional finite surface element on the corneal surface and its bi–dimensional projection on the photography in order to allow an accurate integration of the actual corneal lesion size from the corneal photography. The basic anatomical dimensions of C57BL/6 mice cornea and their variability with age were studied. Geometric lesion of different sizes and locations were created on 6 corneas of 6 different mice, and corneal neovasular lesions were produced in another 6 corneas of 6 mice. In–vivo digitalized slit–lamp pictures were taken at different photographic angles for each lesion and flat mounted cornea were photographed to determining the actual size of corneal lesions. In–vivo measurement of the size of corneal lesions were performed using an improved image analysis software and NIH Image software (NIH Image; Image J, Fredrick, MD) to calculate the apparent lesion size at each angle. Results: The results obtained using the improved software were statistically similar to the actual corneal lesion size and unaffected by the angle of photography. Meanwhile, the conventional method of quantification using imageJ software showed a statistically significant under–estimation of the corneal lesion size, and it is highly affected by the angle of photography. The magnitude of errors of lesion size measurements using the improved software were: –0.21% to 1.46% in the group of central lesions; –1.96% to –0.41% in the group of paracentral lesions; 0.03% to 0.27% in the group of eccentric lesions; –3.59% to –1.70% in the group of mid–peripheral lesions; and –2.29% to 0.13% in the group of peripheral lesions. The magnitude of errors produced by the conventional method using imageJ program were: 12.57% to 12.64% in the group of central lesions; 13.52% to 15.58% in the group of paracentral lesions; 16.95% to 18.30% in the group of eccentric lesions; 21.45% to 25.05% in the group of mid–peripheral lesions; and 32.65% to 43.94% in the group of peripheral lesions. The more peripheral the lesion is located, the larger the magnitude of under–estimation. Conclusions:This new method correct for the parallax induced by the three–dimensional corneal shape, and it permits easy, objective, rapid, and accurate measurements of corneal lesions over extended period of time.

Keywords: cornea: basic science • imaging/image analysis: non–clinical • neovascularization