Classification of Cervical Cancer Using Assembled Algorithms in Microscopic Images of Papanicolaou

In Mexico cervical cancer is the second leading cause of death from malignant neoplasms in women, but this mortality rate has been reduced in recent years thanks to early detection programs as the pap smear test, which is aimed at finding pre-cancerous abnormalities in cells that cover the cervix. The pap smear test is an efficient medical test, but it presents problems at the moment of interpretation under the microscope, due to the large number of cells in the sample and others external factors. In order to solve this disadvantage, computational techniques are used to support the samples classification. In this research we propose to use assembled algorithms to construct a classifier. The database used is from Herlev University Hospital, the data were formulated as a binary classification problem. The results of the experiments (exhaustive search) show that using the combinations of algorithms Bagging+MultilayerPerceptron and AdaBoostM1+LMT is obtained a high percentage of correctly classified instances, 95.74%.