Table 2.

Performance comparison of RS1, RS2, Delta-RS/RS1, C-U-R model radiomics models.

	Training cohort				Validation cohort
	RS1	RS2	Delta-RS/RS1	C-U-R model	RS1	RS2	Delta-RS/RS1	C-U-R model
AUC	0.739	0.863	0.850	0.902	0.748	0.817	0.799	0.885
Accuracy	0.787	0.830	0.806	0.861	0.793	0.829	0.779	0.857
Precision	0.721	0.762	0.712	0.795	0.667	0.731	0.620	0.767
Sensitivity	0.500	0.750	0.795	0.818	0.528	0.722	0.861	0.722
Specificity	0.873	0.835	0.780	0.873	0.856	0.817	0.606	0.904
Recall	0.352	0.546	0.477	0.659	0.389	0.528	0.361	0.639
F1-score	0.473	0.636	0.571	0.721	0.491	0.613	0.456	0.697
Youden Index	0.373	0.585	0.575	0.691	0.384	0.540	0.467	0.626

AUC, the area under the ROC curve; RS, radiomics signature; C-U-R, clinicopathological features, ultrasound features, and RS2.