Computational Detection of Cervical Uterine Cancer

摘要

The aim of this work is to provide a computational method of calculation of parameters to analyze the exfoliative cytology of the cervix or Papanicolaou (PAP). The parameters calculated are those currently used to diagnose cervical uterine cancer, checked by a cytologist through visual examination of a PAP sample, under a microscope. The aim is to determine the risk of cancer, based on the calculation of parameters of the cell nucleus of the sample, such as its size and shape. It is suggested that a low-cost automated system implements a 4-stage process: (1)to obtain a digital image as from a cytological sample; (2)preprocessing it, implementing mathematical operations; (3)to apply an edge detection algorithm to segment the image, separating and identifying the cells from the rest of the sample; (4)to identify a cell nuclei, classify and measure it, to determine whether it corresponds to a healthy or sick nuclei. This process made possible to detect and measure the nucleus of a sample with an efficiency level of more than 96%. This validates the possibility of automating the process of anomaly detection to prevent the cervical cancer, given that it involves great benefits for the patient.