Learning to combine decisions from multiple mammography views

In this paper we address the problem of differentiating between malignant and benign tumors based on their appearance in the CC and MLO mammography views. We describe a two-step classification method that is based on a view-level decision, implemented by a logistic regression classifier, followed by a stochastic combination of the two view-level indications into a single malignant or benign decision. The EM algorithm is used to find the parameters of the proposed model. Our method was evaluated on a standardized digital database for screening mammography (DDSM). Experimental results demonstrate the effectiveness of optimally combining the decisions based on the two views.