Classification of breast tissues in mammographic images in mass and non-mass using mcintosh's diversity index and SVM

Authors:
Péterson Moraes de Sousa Carvalho;Anselmo Cardoso de Paiva;Aristófanes Corrêa Silva
Affiliations:
Applied Computing Group NCA/UFMA, Federal University of Maranhão --- UFMA, São Luís, MA, Brazil;Applied Computing Group NCA/UFMA, Federal University of Maranhão --- UFMA, São Luís, MA, Brazil;Applied Computing Group NCA/UFMA, Federal University of Maranhão --- UFMA, São Luís, MA, Brazil
Venue:
MLDM'12 Proceedings of the 8th international conference on Machine Learning and Data Mining in Pattern Recognition
Year:
2012

Citing 7
Cited 0

Digital Image Processing

Digital Image Processing
Characterization of clustered microcalcifications in digitized mammograms using neural networks and support vector machines

Artificial Intelligence in Medicine
Classification of breast tissues using Moran's index and Geary's coefficient as texture signatures and SVM

Computers in Biology and Medicine
Contourlet-based mammography mass classification using the SVM family

Computers in Biology and Medicine
LIBSVM: A library for support vector machines

ACM Transactions on Intelligent Systems and Technology (TIST)
Proceedings of the Second international conference on Advances in Visual Computing - Volume Part II

ISVC'06 Proceedings of the Second international conference on Advances in Visual Computing - Volume Part II
A new feature descriptor derived from Hilbert space-filling curve to assist breast cancer classification

CBMS '10 Proceedings of the 2010 IEEE 23rd International Symposium on Computer-Based Medical Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper introduces three approaches, which use McIntosh's Diversity index to extract breast tissue features from mammographic images, for later classification, through Support Vector Machine (SVM), into mass and non-mass. In order to implement the diversity index, it is necessary to define the element that will represent the species in the image. So, in the first approach, the intensities of the pixels of the image are treated as species, and the texture statistic used is the histogram. Considering the spatial relations of direction and distance between pixels, we adopted a second approach, using GLCM as texture statistic, where the species are represented by pairs of pixels, and the third approach, using GLRLM as texture statistic, where the species are represented by gray level run lengths. We achieved an accuracy of 60.25% with the first approach, 99.00% with the second one and 99.75% with the third one.