Extraction and Separation of Overlapped Squamous Cell Cytoplasm with Disjoint Nuclei in Cervical Pap Smear Image

Abstract

The automated analysis of Papanicolaou (Pap) smear images is a challenging issue due to the occlusion of cells. The feature extraction and interpretation depend on the accuracy of cell segmentation. This is important to find the difference between normal and abnormal cells. However, most of the time complications are due to overlapping in the Pap smear cells. The overlapping can be in nuclei regions or cytoplasmic regions or sometimes both the regions. There are two types of cells found in Pap smear images – Columnar and Squamous cells. Most of the time abnormalities are found in squamous cells which in turn may lead to cervical cancer. Hence, this work concentrates on the separation of squamous cells in which cytoplasmic regions are overlapped with disjoint nuclei regions. The main objective of this work is to identify the intersecting points where cytoplasmic regions of two cells meet, called Concavity points, which are calculated using a scope of points along the cytoplasmic boundary of the cells, and separating the cells along with these concavity points.