Feature Analysis of Normal Epithelial Cervical Cell Characteristics in Pap Smear Images

We evaluate the characteristic of the normal epithelial cervical cell in Pap Smear images, using feature analysis. The evaluation affects the determination of proper pap smear image determination. This study aims to analyze the performance of feature selection on data classification and discovering features which significantly affect the classification of the normal epithelial cervical cell. Feature selection process has been done to 54 features in the nuclei area and the cytoplasm of the cervical epithelial cell, using Feature Subset Selection. Furthermore, we compare the performance of two classification methods: K-Nearest Neighbors (KNN) and Backpropagation. Both methods resulting in the same 12 features to differentiate between normal cervical cells. The classification accuracies for both methods are 92.29% for KNN and 91.51% for Backpropagation.