@Article{oncologie.2020.013660, AUTHOR = {Wan Azani Mustafa, Afiqah Halim, Khairul Shakir Ab Rahman}, TITLE = {A Narrative Review: Classification of Pap Smear Cell Image for Cervical Cancer Diagnosis}, JOURNAL = {Oncologie}, VOLUME = {22}, YEAR = {2020}, NUMBER = {2}, PAGES = {53--63}, URL = {http://www.techscience.com/oncologie/v22n2/40105}, ISSN = {1765-2839}, ABSTRACT = {Cervical cancer develops as cells transformation in the cervix of a female that connects the uterus to the vagina. This cancer may impact the columnal epithelial cells of the cervix and therefore can be expanded to the lymphatic and circulatory system (metastasize), sometimes the kidneys, liver, prostate, vagina, and rectum. Many of the cervical cancer patients survived by taking early prevention by undergoing a Pap Smear Test. However, the result of the test usually takes a few weeks which is extremely time-consuming especially at the government hospital. The purpose of this research was to study the detection and classification method of the Pap Smear image to resolve the timeconsuming issues and support better system performance to prevent low precision result of the Human Papilloma Virus (HPV) stages. A few studies were considered which features the cell image databases to classify cervical cancer according to its type. Besides, the classification system and the performance of the preceding papers that had been considered include a few features found in the cell images. Those features were the size of the cells, the shape of the cells, the colour, Region of Interest (ROI) and overlapped cell nuclei. The other existing design methods being considered were the Deep Convolutional Neural Network (CNN) and the Artificial Neural Network (ANN). These findings technique showed the highest percentage of the system accuracy, precision, and specificity that might be excellent for further analysis. The research limitation was the method of how the numerous image databases needed to be processed and classified one at a time. None of these articles stated whether they had found the way to compute more images at once. The aim of the study was to review the previous paper in order to define the feature datasets that needed to be considered. The features were important in designing a new classification method and increasing the performance of the systems. The features included the nucleus shape, diameter and surface areas, colour and luminosity of the cell datasets, the region of the nucleus, design and image resolution. In this paper, an extensive analysis was studied for cervical cancer classification techniques. As expected from the outcome, the study of the feature database, the classification method and the system performance were reviewed deeper for further assessments.}, DOI = {10.32604/oncologie.2020.013660} }