False Positive Reduction in Breast Mass Detection Using Two-Dimensional PCA

In this paper we present a novel method for reducing false positives in breast mass detection. Our approach is based on using the Two-Dimensional Principal Component Analysis (2DPCA) algorithm, recently proposed in the field of face recognition, in order to extract breast mass image features. In mammography, it is well known that the breast density measure is highly related to the risk of breast cancer development. Hence, we also propose to take advantage of a previous breast density classification in order to increase the overall breast mass detection performance. We test our approach using a set of 1792 RoIs manually extracted from the DDSM database. Moreover, we compare our results with several existing methods. The obtained results demonstrate the validity of our approach, not only in terms of improving the performance but being a generalizable, simple, and cost-effective approach.