BI-RADS Features-Based Computer-Aided Diagnosis of Abnormalities in Mammographic Images

Abstract

In the analysis of digital or digitized mammographic images, a requirement is to learn to separate benign abnormalities from malignant ones. Such an activity could form part of a computer-aided diagnosis (CAD) tool. We present a CAD study of mass and calcification lesions found in digital database of screening mammography (DDSM) using BI-RADS-based features to demonstrate the performance of feature elimination-based support vector machines as the classification technique. It is shown that using only a subset of the available set of features is shown to significantly better classify abnormalities. Furthermore, it is also shown that CAD of same-institutional mammograms produces higher classification accuracy in general compared to that of cross-institutional mammograms.