Skip to main content
. 2024 Sep 17;7(3):e70001. doi: 10.1002/jsp2.70001

FIGURE 3.

FIGURE 3

Confusion matrices (left) and receiver operating characteristic (ROC) curves (right) of MAIA models regardless of the presence of crosslinks. Range of the area under the ROC curve (AUC): (A) AP model, 0.93–1; (B) Lat model, 0.95–1; (C) Concat model, 0.92–1; (D) Merge model trained on dual images, 0.89–1; (E) Merge model trained on AP images, 0.83–1; and (F) Merge model trained on lateral images, 0.93–1. The x‐ and y‐axis in the confusion matrices represent the true labels and the predicted labels, respectively. Darker blue in the confusion matrices represents higher values. Lines are colored to indicate the following: blue, ROC curve of A‐Spine; red, ROC curve of Armstrong; green, ROC curve of CDH; light blue, ROC curve of Expedium; lavender, ROC curve of Gezen; yellow‐green, ROC curve of NOVA; dark blue, ROC curve of Xia 3; shocking pink dotted line, micro‐average ROC curve; oriental blue dotted line, macro‐average ROC curve.