A comparison of multi-resolution and multi-orientation for breast cancer diagnosis in the full-field digital mammogram

This paper presents a comparative study between Steerable Pyramid, Curvelet, Contourlet, and Gabor for breast cancer diagnosis in the full-field digital mammogram. Using multi-resolution and multi-orientation analysis, mammogram analysis images are decomposed into different resolution levels and orientations. A set of the biggest coefficients from each scale and orientation is extracted. Then a supervised classifier system based on Support Vector Machine is constructed. The experimental results gave a 98.33% classification accuracy rate, which indicate that Steerable Pyramid transformation is a promising tool for analysis and classification of full-field digital mammograms.