I. Introduction
Breast cancer is one of the most frequent leading causes of cancer deaths in women. The American Cancer Society (ACS) estimates that about 184,450 new breast cancer cases are expected to be diagnosed in 2008 [1]. Therefore, the early detection is a main factor to reduce deaths of the disease. Mammography, which reveals the pronounced evidence of abnormality in breast, is currently the most effective tool for early detection of breast cancer. However, it is difficult to interpret a mammography as its sensitivity is seriously affected by image quality and radiologist's experiences. Independent double reading by two radiologists is introduced in screening routine to improve the accuracy of diagnosis. Though it could improve the sensitivity of diagnosis, the high cost is unacceptable in practical applications. Therefore, Computer-aided detection (CAD) schemes have been developed and acknowledged to assist radiologists in improving the accuracy of diagnosis. Masses and calcifications are two primary signatures of abnormity in mammograms. Existing research results show masses are more difficult to recognize because of their abundant appearances and ambiguous margins than calcifications, and thus, mass detection is a challenging problem [2] [3].