Skip to main content
. 2020 Mar;6(1):23–33. doi: 10.18383/j.tom.2019.00021

Figure 7.

Figure 7.

Recurrence classification from features calculated from pre-RT images receiver operating curve (ROC) curves for prediction accuracy using neural networks and support vector machines (SVMs) for 3 regions examined. The best predictive power is found from features in the peritumoral area (neural network [NN] AUC: 0.78), followed by the tumor and tumor combined with the peritumoral area (NN AUC: 0.68 vs 0.96). The NN outperforms the SVM for all regions.