Abstract
Purpose: :
To assess the validity of an automated way to count basal epithelial cell compared to manual count of basal epithelial cells.
Methods: :
20 optimal images of basal epithelial cells obtained using confocal microscopy (ConfoScan3, Nidek Technologies, Italy) were identified and imported into ImageJ (image processing software). The images were then cropped to a size of 128x128 pixels (ROI) and background subtraction was performed twice on these images (rolling ball size 50). The number of basal epithelial cells within the ROI, were then automatically counted using the plugin "ITCN" (image-based tool for counting nuclei). This procedure was performed twice for each ROI. The same ROI was then used for the manual cell count. The manual count was performed twice for each ROI by one trained person. Statistical significant differences between the two methods were determined using repeated measures ANOVA, and concordance correlation coefficients (CCC) were computed to determine test-retest and method comparison concordance.
Results: :
Analysis showed no statistical significant difference (p<0.05) between the manual (26.7 ± 1.5) and the automated cell count (26.1 ± 2.5). Also, no statistical significant differences were found between the two trials for each method (p<0.05). The CCCs were 1.0, 0.7 and 0.6 for automated, manual and automated/manual, respectively (all p<0.05).
Conclusions: :
The results suggest that there is no difference in counting the cells using the automated way compared to the manual count.
Keywords: cornea: epithelium • imaging/image analysis: clinical • image processing