Output details

Toward breast cancer diagnosis based on automated segmentation of masses in mammograms

Breast cancer is an increasingly growing problem in the world; effective mass-screening can only be achieved with computer-aided techniques, which currently require experts identifying breast-mass boundaries in mammograms. This paper overcomes this limitation by inventing automated segmentation methods and novel characterization features for the classification of breast-masses into either benign or malignant category. Simplified versions of mass contours generated through rigorous studies were employed to extract features for characterization of breast-masses. These novel techniques were shown to make significant contributions to accurate diagnosis of breast masses in screening programs. This highly original research already received 40 citations (Google Scholar).