Table 3.

The performance of the deep learning model on both retrospective and prospective dataset.

	Sensitivity (%)	Specificity (%)	Accuracy (%)	PPV (%)	NPV (%)
Internal
Retrospective testing
Per patient	100	93.55	95.24	84.62	100
Per image	94.34	99.16	98.85	88.37	99.61
Prospective testing (per patient)	100	81.82	92.59	88.89	100
External
Retrospective testing (per patient)	98	94	96	94.23	97.92

PPV positive prediction value; NPV negative prediction value.