832509.fig.001
Figure 1: Inputs to the network are lesion descriptors and family history of the patient. Nodes at each layer are connected to the nodes at the succeeding layer by weighted arcs. Each hidden node in the first hidden layer performs a nonlinear weighted sum of all input values. The outputs of the last hidden layer are then similarly combined to the output layer. The single output value shows the probability of the lesion being malignant.