Figure 1.

Transfer learning networks, feature embeddings, and baseline results. (A) ILSVRC network models utilized, with their top 1% and top 5% accuracy in ILSVRC competition noted. (B) Example CT and MRI images for both ACP and NOTACP. (C) ROC (left) and AUC (right) values for all twelve networks and both imaging modalities (CT top, MRI bottom). The diagonal dashed line represents performance of a random guess.