Título: Classificação de Tumores e Massas de Mama Utilizando um Comitê de Perceptrons de Múltiplas Camadas
Autores: Silva, Leandro Augusto da; Hernandez, Emilio del Moral; Rangayyan, Rangaraj M.
Resumo: This paper addresses a new approach using the committee machine to classify masses in mammograms as benign or malignant. Three shape factors and three measures of edge sharpness were used for the classification of 37 regions of interest (ROIs) related to benign masses and 20 ROIs of malignant tumors. The committee machine is a group of classifiers used to resolve a difficult task. In this work, we used a group of mulit-layer perceptrons (MLP) as a committee machine, and the classification results were realized by combining all of the classifiers’ responses. The classifier’s performance was evaluated by the area under the receiver operating characteristics (ROC) curve (Az). The Az result for the committee machine was compared with the Az results of the MLP and single-layer perceptrons (SLP) neural networks. With the shape features plus edge-sharpness features, the Az result for the committee machine (0.88) was significantly better than that for the MLP (0.76) and SLP (0.54).
Código DOI: 10.21528/CBRN2005-195
Artigo em PDF: CBRN2005_195.pdf
Arquivo BibTex: CBRN2005_195.bib