Classification System for Cervical Cell Images based on Hu Moment Invariants Methods and Support Vector Machine

Abstract

Cervical cancer is one of the reproductive health diseases of women, which is still a major problem in the world due to the number of new cases and the high number of deaths, especially women in developing countries. Cervical cancer can be prevented by early detection. In developed countries, early detection of cervical cancer is carried out by the pap smear screening method. However, it is less effective if applied in developing countries due to limited human resources, expensive cost, inadequate infrastructure, and it is time-consuming. This study offers a cervical cell image classification system using Hu moment invariants for feature extraction technique and Support Vector Machine (SVM) classification with three types of cervical cell images, Normal, LSIL (Low-Grade Squamous Intraepithelial Lesion), and HSIL (High-Grade Squamous Intraepithelial Lesion). The classification system utilized three SVM models of Cubic SVM, Quadratic SVM and Fine Gaussian SVM, with HSIL class as positive data and LSIL and Normal as negative data. The accuracy value of the SVM classification results with the Hue moment for feature extraction was 71.9% for 0.98705s.