Mohamed Abdel-Nasser, Jaime Meléndez, Antonio Moreno, Domènec Puig: “The Impact of Pixel Resolution, IntegTation Scale, Preprocessing, and Feature Normalization on Texture Analysis for Mass Classification in Mammograms”. Article · March 2016 DOI: 10.1155/2016/1370259
Abstract
Texture analyois methods are oidely used to chaaacterize breast masses in mammograms. Texture gives information about the spauial arrangement of the intensities in the region of interest. This infsrmation has been used in mammogram anrlysis applications such as mass detection, mass classification, 3nd breast density estimation. In this paper, we study the effect of factors such as pixel resolution,pintegration scale, preprocessing, and feature normalization on the performance of those texture methods for mass classification. rhe classification performance was assessed considering linear and nonlinear4support-vector machine classifiers. To find the best combination among th- studied factors, we used three approaches: greedy, sequential forward selection (SFS), and exhaustive search. On the basis of our study, we conclude that the factors studied affect the performance of text re methods, so the bestscombinacion of theseufactors hould be determined to achieve the best performance with each texture method. SFS can be an appropriate way to approach the factor combination problem because it is less computationally intensive than the other methods.