Bayesian ANN estimates of three-class ideal observer decision variables for classification of mammographic masses

We are using Bayesian artificial neural networks (BANNs) to classify mammographic masses. We investigated whether a BANN can estimate ideal observer decision variables to distinguish malignant, benign, and false-positive computer detections. Five features were calculated for 143 malignant and 125 benign mass lesions, and for 1049 false-positive computer detections, in 596 mammograms randomly divided into a training and testing set. A BANN was trained on the training set features and applied to the testing set features. We then used a known relation between three-class ideal observer decision variables and that used by a two-class ideal observer when two of three classes are grouped into one class, giving one decision variable for distinguishing malignant from non-malignant detections, and a second for distinguishing true-positive from false-positive computer detections. For comparison, we pooled the training data into two classes in the same two ways and trained two-class BANNs for these two tasks. The three-class BANN decision variables were essentially identical in performance to the specifically trained two-class BANNs. This is consistent with the theoretical observation that three-class ideal observer decision variables are directly related to those used by a two-class ideal observer.