Reliability of nucleus-to-cell and nucleus-to-cytoplasm calculations for conjunctival impression cytology specimens.

Purpose: To assess the outcome of two different nucleo-cytoplasmic ratio calculations in relation to the number of cells measured from conjunctival impression cytology (CIC) images.
Methods: CIC from the exposed nasal bulbar conjunctiva surface was undertaken on 40 young adults using a Millcell(®)-CM filter; the filters were stained with Giemsa, colour images taken of the cells in a monolayer at 200 X magnification, graded by the Nelson scale and then a projection overlay method used to outline the cell and nucleus borders. Nucleo-cytoplasmic ratios were calculated based on literal interpretation of the term (i.e. N/CYT = nucleus area/[cell area - nucleus area]) as well as based on cell area (i.e. N/CELL = nucleus area/cell area).
Results: A range of N/CYT or N/CELL values are expected for all cells, regardless of squamous metaplasia grade, with slightly greater heterogeneity expected from N/CYT calculations. The difference between the two calculations gets progressively less as the squamous metaplasia grade increases. For all grades of cells, these ratios could be determined to within approximately ±1% if adequate numbers of cells were measured (i.e. 75, 65, 50 and 35 for grade 0 to grade 3 cells respectively). If fewer cells are measured, then the variability in N/CELL or N/CYT calculations can be expected to increase.
Conclusions: In assessments of nucleo-cytoplasmic ratios, it is important that the calculation used be specified and that the numbers of cells measured be provided.