Avaliação de Algoritmos de Classificação de Padrões na Detecção de Câncer do Colo do Útero

Abstract

The main goal of this study is to empirically evaluate linear and nonlinear pattern classifiers in the detection of cervical cancer. For this purpose, this paper reports a comprehensive performance comparison study involving 4 different classification methods, namely: linear least squares classifier, extreme learning machine, suport vector machine and quadratic Gaussian classifiers. All the classifiers are evaluated on a benchmarking database generated from pap-smear image analysis. The results of the experiments indicate that the linear classifiers perform as good as powerful nonlinear ones, with the advantage of demanding much lower computational resources. Palavras-chave: Exame Papanicolau, câncer cervical, classificação de padrões, SVM, ELM.