During the Pap smear examination, the cytopathologist follows a protocol defined by the Bethesda system to analyze cell morphology to perform a lesion detection.
Thus, the development of a semi-automatic computational tool capable of increasing the reliability of the Pap Smear test results involves detecting the cell nuclei and cytoplasm.
We have been developing techniques for detecting and classifying specific regions of the cervical cells, especially the nucleus. We investigate eight traditional machine learning methods to perform a cervical cell classification [1]. We propose a hierarchical classification methodology for computer-aided screening of cell lesions, which can recommend fields of view from the microscopy image based on the nuclei detection of cervical cells. We evaluate the performance of algorithms against the Herlev and CRIC searchable image database.
Our new proposal involves deep convolutional neural networks for cervical cell lesion classification. It proposes an ensemble of the three best architectures to classify cervical cancer based on cell nuclei [3]. The dataset used in the experiments is available in the CRIC Searchable Image Database. Considering the precision, recall, F1-score, accuracy, and sensitivity metrics, the proposed ensemble improves previous methods from the literature for two- and three-class classification. We also introduce the results for the six-class classification.
This research proposes a workload-reducing algorithm for cervical cancer detection based on analysis of cell nuclei features within Pap smear images.
Team: Débora, Marcone, Mariana, Daniela, Fátima, Claudia and Andrea.
Publication
[1] D.N. Diniz, M. T. Rezende, A. G. C. Bianchi, C. M. Carneiro, D. M. Ushizima, F. N. S. de Medeiros, and M. J. F. Souza. A Hierarchical Feature-Based Methodology to Perform Cervical Cancer Classification. Applied Sciences 2021, 11(9), 4091.
[2]D. N. Diniz, R. F. Vitor, A. G. C. Bianchi, S. E. D. Silva, C. M. Carneiro, D. M. Ushizima, F. N. S. de Medeiros, and M. J. F. Souza. A ensemble method for nuclei detection of overlapping cervical cells. Submitted to Expert Systems and Applications.
[3] D. N. Diniz, M. T. Rezende, A. G. C. Bianchi, C. M. Carneiro, E. J. S. Luz, G. J. P. Moreira, D. M. Ushizima, F. N. S. de Medeiros, and M. J. F. Souza. A Deep Learning Ensemble Method to Assist Cytopathologists in Pap Test Image Classification. Submitted to Journal of Imaging.
