Table 4.

Performance of clinical, radiomics, deep learning radiomics, and combined models.

Model	Cohort	ACC	AUC (95% CI)	SEN	SPE	PPV	NPV	F1-score	Threshold
Clinical	Training	0.852	0.804 (0.750–0.857)	0.943	0.591	0.868	0.783	0.904	0.656
Radiomics	Training	0.882	0.943 (0.917–0.970)	0.886	0.873	0.952	0.727	0.918	0.787
DLR	Training	0.915	0.970 (0.955–0.986)	0.911	0.927	0.973	0.785	0.941	0.842
Combined	Training	0.918	0.972 (0.958–0.987)	0.917	0.918	0.970	0.795	0.943	0.877
Clinical	Validation	0.856	0.820 (0.737–0.902)	0.927	0.628	0.888	0.730	0.907	0.738
Radiomics	Validation	0.917	0.912 (0.852–0.973)	0.964	0.767	0.930	0.868	0.946	0.431
DLR	Validation	0.894	0.940 (0.896–0.984)	0.912	0.837	0.947	0.750	0.929	0.566
Combined	Validation	0.889	0.942 (0.902–0.982)	0.905	0.837	0.947	0.735	0.925	0.537

DLR, Deep learning radiomics; AUC, area under the curve; CI, confidence interval; PPV, positive predictive value; NPV, negative predictive value;ACC, Accuracy;SEN, Sensitivity; SPE,Specificity.