A Semi - Automated Morphometric Assessment of Nuclei in Pap Smears Using Imagej

Abstract

BACKGROUND AND PURPOSE: Carcinoma of cervix is the fourth commonest malignancy in women. Its incidence is progressively falling due to the routine use of Pap smears to detect precancerous lesions. However, routine Pap smear examination is time consuming and, as it is based on descriptive morphological assessment, false positive or negative reports are likely to occur. Using morphometric techniques, several attempts have been made to improve the accuracy of reports. In the present study, we have used Image morphometric software and some of its plugins to create a macro to analyse large number of cells at a time. MATERIALS AND METHODS: Using Image and three of its plugins, namely, BEEPS, Kuwahara filter and Mexican Hat filter, we created a macro to morphometrically analyse normal, reactive and neoplastic Pap smears. We also compared the macro measurements with manual measurements. RESULTS AND CONCLUSIONS: Results obtained with macro showed strong positive correlation with manual measurement. Although the neoplastic nuclei were on an average larger than reactive/normal nuclei, there was considerable overlap. More than the enlargement, anisonucleosis (variability in the size) appeared to be a better indicator of neoplasia. The macro that we developed works rapidly and gives results comparable to manual measurements provided the smears and the photographs are technically acceptable.