Muqing Lin1, Ke Nie1, Jeon-Hor Chen1,2, Orhan Nalcioglu1, Min-Ying Lydia Su1
1Tu & Yuen Center for Functional Onco-Imaging, University of California, Irvine, CA, USA; 2Department of Radiology, China Medical University, Taichung, Taiwan
The purpose is to evaluate the reduction of vessel contamination in the hot-spot search of breast lesions on DCE-MRI. This is particularly important for development of automated CAD that identified lesions and characterizes the suspicious impression. The blood vessels show strong contrast enhancements and the wash-out pattern, which is a strong malignancy feature. The initial hot spot search was performed using a kernel of 3x3 pixels throughout the entire breast. The vessels were identified using a computer algorithm-based method based on 2D and 3D. We have shown that our algorithm can successfully exclude vessels and reduce false positive detection.