Multilayer Perceptron for squamous cell nuclei detection in Pap smear tests for cervical cancer diagnosis using local features

Cervical cancer is the fourth most frequently diagnosed cancer among women worldwide, and it is a leading cause of death in women. The diagnosis is performed manually by analyzing the cells in Pap smear tests, this causes diagnoses to be delayed and results in a high rate of false positives. To address this issue, a Multilayer Perceptron (MLP) artificial neural network model is proposed, utilizing local features to automatically classify squamous cell nuclei. For model training, two public datasets were used: CRIC and SIPaKMeD. Unlike many of the reported methods of the state of the art, our proposal learns from the content of whole images, achieving an accuracy of 0.9526 and 0.9397 on the CRIC and Sipakmed datasets, respectively. Our proposal provides a simple approach while maintaining performance comparable to that of complex models reported in the literature.