• Korean Journal of Clinical Laboratory Science
• /
• v.38 no.3
• /
• pp.203-207
• /
• 2006

The purpose of this study was to develop discriminant analysis models for predicting cervical normal/dysplasia case diagnoses using cytometric features derived from the digital image analysis of cell monolayers. The database consisted of 19 cases diagnosed either as normal (n=5), moderate dysplasia (n=7), severe dysplasia (n=7) on monolayer preparations. We studied the nuclear and cytoplasmic characteristics of cells in the normal, moderate dysplasia and severe dysplasia on cervical samples. The morphometric parameters selected for the analysis were nuclear/cytoplasmic ratio and the nuclear variations measured by image analysis on normal and precancerous lesions of cervical smears; several shape factors; area; perimeter; maximal, minimal and equivalent circle diameters. The results showed that the dysplasia samples exhibited changes in both cellular and nuclear form and size but lacked substantial differences in the tumor grades. The coefficient of nuclear variation is as follows to normal cell $21.8{\pm}3.2%$, moderate dysplasia $33.5{\pm}6.1%$, severe dysplasia $27.7{\pm}5.8$ of cervical smears.