Figure 5.

The 5-fold cross-validation results for the ROC curves and AUC values of the different frameworks. The results of different cross-validation models are shown in different colors. A: RetinaNet [25]. B: Liao et al [15]. C: Duan et al [16]. D: Bi-input+RetinaNet+C-LSTM. The ROC curves of fold 0 and fold 2 in graph C overlap, and the ROC curves of fold 1 and fold 4 in graph D overlap. AUC: area under the curve; C-LSTM: convolutional long short-term memory; ROC: receiver operating characteristic.