A Textural Approach for Recognizing Architectural Distortion In Mammograms

Breast cancer is considered as the mostimportant cause of death among women. Architecturaldistortions are very important signs of breast cancer and earlydetection of them is a rewarding work. In this paper wepropose a method to recognize architectural distortion fromnormal parenchyma. In our proposed method, appropriatefeatures are extracted by the analysis of oriented textures withthe application of orientation component of recent the state-ofthe-art local texture descriptor called Monogenic BinaryCoding (MBC). In addition, we transform Region of Interests(ROIs) to polar coordinates in order to highlight some specificpatterns in mammograms. Various classifiers are used over aset of mammograms from Digital Database for ScreeningMammography (DDSM). The results show that proposedmethod is very encouraging. The best performance achieved is91.25% in terms of the average accuracy using the NearestNeighbor classifier.

Keywords:

breast cancer , mammogram , architectural distortion , local texture descriptor , monogenic binary coding distortion , local texture descriptor , monogenic binary coding