Resumen

RODRIGUEZ VAZQUEZ, Solangel; MARTINEZ BORGES, Andy Vidal  y  LORENZO GINORI, Juan Valentín. Cervical cell classification by means of the KNN algorithm using nucleus’ features. Rev cuba cienc informat [online]. 2016, vol.10, n.1, pp. 95-109. ISSN 2227-1899.

ABSTRACT The Pap test is a test of gynecological screening that allows appreciating changes in the morphology of the cells of the cervix. This study can alert on such frequent pathologies in women as cancer of the cervix. The analysis of these kinds of images is important in the generation of diagnostic and the investigations that carried out, so that developing new techniques that made a practical analysis of the samples is necessary. Similarity search is one of the most common procedures in problems involving processing of data, an alternative to solve this problem is the kNN search (k-Nearest Neighbors). In this paper, the kNN classifier was used together with a specific distance function, to provide a solution to the real problem associated with the classification of cervical cells in normal and abnormal classes, the features used for classification were in this case based solely on information extracted from the nuclei region. From the study, among the manhattan distance, Euclidean and Mahalanobis and considering measures for evaluating F, AUC, negative predictivity and H-mean was found that manhattan performed well holding 97.1% values of AUC. The results indicate a reduction compared to the rate of false negative Pap test. H-mean with the purpose of comparing the results of other investigations regarding kNN, obtaining 92.33% with regard thereto.

Palabras clave : cervical cell; classification; kNN; nuclei; Pap test.

        · resumen en Español     · texto en Español     · Español ( pdf )