Use of a New Model for Signal Representation, as a Feature Extractor, in Content-Based Medical Image Retrieval

• João L. Fernandes UFF
• Cristina N. Vasconcelos UFF
• Giovanni C. Carelli UFF

Abstract

The recovery of similar items from medical images databases can promote the acquisition of important knowledge to aid the diagnosis. The recovery of images by similarity involves the extraction of representative features of the images, which is not a trivial task, since there is usually a large number of factors that must be considered to simulate the human perception. Such techniques are known in literature as CBIR (Content-based Image Retrieval). In this paper, we propose studying a new model for signal representation, as features extractor, aiming at developing metrics of similarity between medical images.

References

Castañon, C. A. B. (2003) “Recuperação de imagens por conteúdo através de análise multirresolução por Wavelets”, Dissertação de Mestrado, ICMC-USP, São Carlos.

Fernandes, J. L. (2006) “Shape from shading: Novas abordagens a partir de movimento e de foco”, Tese de Doutorado, Universidade Federal Fluminense, Niterói.

Silva, L. A. (2009) “Categorização de Imagens Médicas baseada em Transformada Wavelet e Mapas Auto-Organizáveis”, Tese de Doutorado, USP, São Paulo.

Suckling, J. (Organiser) (2010) site “The mini-MIAS database of mammograms” [link], Último acesso: Agosto/2010.

Torreão, J. R .A. e Victer, S. M. C. (2010) “A Transformada de Gabor Sintonizada”, Relatório técnico apresentado ao Instituto de Matemática Pura e Aplicada (IMPA), Rio de Janeiro.

Torreão, J. R. A., Fernandes, J. L. and Victer, S. M. C. C. (2009) “A Model for Neuronal Signal Representation by Stimulus-Dependent Receptive Fields” In: 19th International Conference on Artificial Neural Networks (ICANN 2009), Limassol.