Applied Mathematics & Information Sciences

Author Country (or Countries)



In this paper, fusion of texture features to improve classification accuracy by false positive reduction in mammograms is proposed. The method uses texture features obtained from completed local binary pattern (CLBP) and grey level texture features obtained from the Curvelet sub-bands. In the current experiments, mass and normal patches were obtained from Mammographic image analysis Society (MIAS) and Image retrieval in medical applications (IRMA) datasets for mammograms. Texture features from both methods are combined together to obtain the feature fusion matrix. Then Nearest neighbor classifier was used for classification to evaluate the individual as well as enhanced features obtained from CLBP and curvelet. The classifier produces a classification accuracy of 96.68% with 98.9% sensitivity and the false positive (FP) rates drop by 40% and 78% respectively for the enhanced features as compared to the original results produced by both methods. The experimental results suggest that fusion of features improves the performance of the system and is statistically significant.