Breast cancer mostly arises from the glandular (dense) region of the breast. Consequently, breast density has been found to be a strong indicator for breast cancer risk. Therefore, there is a need to develop a system which can segment or classify dense breast areas. In a dense breast, the sensitivity of mammography for the early detection of breast cancer is reduced. It is difficult to detect a mass in a breast that is dense. Therefore, a computerized method to separate the existence of a mass from the glandular tissues becomes an important task. Moreover, if the segmentation results provide more precise demarcation enabling the visualization of the breast anatomical regions, it could also assist in the detection of architectural distortion or asymmetry. This study attempts to segment the dense areas of the breast and the existence of a mass and to visualize other breast regions (skin-air interface, uncompressed fat, compressed fat, and glandular) in a system. The graph cuts (GC) segmentation technique is proposed. Multiselection of seed labels has been chosen to provide the hard constraint for segmentation of the different parts. The results are promising. A strong correlation (r = 0.93) was observed between the segmented dense breast areas detected and radiological ground truth.
The treatment and therapy to be administered on breast cancer patients are dependent on the stage of the disease at time of diagnosis. It is therefore crucial to determine the stage at the earliest time possible. Tumor dissemination to axillary lymph nodes has been regarded as an indication of tumor aggression, thus the stage of the disease. Neural networks have been employed in many applications including breast cancer prognosis. The performance of the networks have often been quoted based on accuracy and mean squared error. In this paper, the performance of hybrid networks based on Multilayer Perceptron and Radial Basis Function networks to predict axillary lymph node involvement have been investigated. A measurement of how confident the networks are with respect to the results produced is also proposed. The input layer of the networks include four image cytometry features extracted from fine needle aspiration of breast lesions. The highest accuracy achieved by the hybrid networks was 69% only. However, most of the correctly predicted cases had a high confidence level.