Cervical Cancer Detection Using Single Cell and Multiple Cell Histopathology Images: Do You Have a Subtitle? If So, Write It Here

Abstract

Cervical cancer is the second most common cancer in females in India. A Pap smear screening is most efficient and prominent to detect the abnormality in cells. Pap smear test is time-consuming and sometimes gives the wrong result by human experts. In India, a shortage of pathologist is there in rural areas. Automated systems using image processing and machine learning techniques help the pathologist to take correct decisions. In this paper, two data sets are generated from one pathologist center. The first data set contains 300 single cells and the second contains 50 multiple cell images for the validation of work. In a single cell, nucleus and cytoplasm both are extracted from the cell, but in multiple cells, only the nuclei are extracted due to overlapping of cells. Edges have been enhanced by sharpening function, and the multi-threshold values and morphological operations have been used for the segmentation of cell. Shape-based features have extracted from a multiple cell and single cell images. Support Vector Machine (SVM) and Artificial Neural Network (ANN) is applied to improve the performance of classification using 10 fold cross-validation.