A global approach for detecting mass in Digital Mammograms
Annamalai Nagar · 2012
This paper proposes some image processing methods for detecting breast masses in mammograms. This approach is global in the sense the breast region along with the background and pectoral muscle is used as input for detecting mass. First, the mammogram is preprocessed for noise reduction and image enhancement. We introduce a thresholding method to obtain rough region of interest. Connected component labeling is then applied to the binarized image to get discrete regions or objects. The objects are subjected to various yes or no criterian which narrows down the region that are most likely to be the mass. The proposed work is done using the Mammography Image Analysis Society (MIAS) database. This work shows that connected component labeling can be used to detect masses in digital mammograms effectively.