Computer-Aided Mass Detection on Digitized Mammogram

In this paper, a segmentation method for detection of suspicious masses in digitized mammograms devel-oped by enhancement and adaptive threshold method is introduced. The algorithm consists of the following steps: 1 - Preprocessing of the digitized mammograms including; image resizing, reduction of the dimension of the image to be processed through the identifica-tion of the region of interest as candidate for massive lesion through breast region extraction by flood-fill operation, 2- Image enhancement using linear trans-formation and subtracting enhanced image from the original image. 3- Characterization of the region of interest by extracting the features. 4- Local adaptive threshold of images obtained from step one, for seg-mentation of mass areas. The proposed method was evaluated on mammograms from the BI-RADS data-set and a local database. The detected regions have been validated by comparing them with the radiolo-gists’ hand-sketched boundaries of real masses. The algorithm exhibits a sensitivity of 87% for mass de-tection when the preprocessing step is applied.