Table 1.

Diagnostic performance of seven-way classifiers and radiologists in the validation set.

	Value (95%CI)			McNemar's P-valuea	Delong P-value
	Model A	Model B	Reader consensus	Model A vs. Reader	Model A vs. Model B
Cyst
AUC	0.986 (0.960, 1.000)	0.970 (0.932, 1.000)			0.147
Sensitivity, %	89.5 (75.2, 97.1)	89.5 (75.2, 97.1)	94.7 (82.3, 99.4)	0.688
Specificity, %	96.9 (93.0, 99.0)	96.9 (93.0, 99.0)	98.2 (94.7, 99.6)	0.727
Hemangioma
AUC	0.944 (0.897, 0.991)	0.936 (0.886, 0.986)			0.66
Sensitivity, %	82.6 (68.6, 92.2)	87.0 (73.7, 95.1)	89.1 (76.4, 96.4)	0.508
Specificity, %	95.5 (90.9, 98.2)	93.6 (88.5, 96.9)	98.1 (94.5, 99.6)	0.344
FNH
AUC	0.987 (0.934, 1.000)	0.936 (0.824, 1)			0.102
Sensitivity, %	100 (66.4, 100)	66.7 (29.9, 92.5)	88.9 (51.8, 99.7)	1.000
Specificity, %	95.8 (92.0, 98.2)	94.8 (90.63, 97.5)	100 (98.1, 100)	0.008
Benign nodules
AUC	0.914 (0.787, 1.000)	0.865 (0.711, 1.000)			0.139
Sensitivity, %	77.8 (40.0, 97.2)	66.7 (29.9, 92.5)	55.6 (21.2, 86.3)	0.500
Specificity, %	99.5 (97.1, 100)	99.5 (97.1, 100)	98.4 (95.5, 99.7)	0.625
HCC
AUC	0.925 (0.871, 0.978)	0.879 (0.813, 0.9452)			0.137
Sensitivity, %	87.2 (74.3, 95.2)	74.5 (59.7, 86.1)	87.2 (74.3, 95.2)	1.000
Specificity, %	91.6 (86.0, 95.4)	86.4 (79.9, 91.4)	94.8 (90.0, 97.7)	0.267
Metastatic malignancy
AUC	0.897 (0.828, 0.966)	0.841 (0.758, 0.923)			0.039
Sensitivity, %	59.6 (42.1, 75.3)	40.5 (24.8, 57.9)	89.2 (74.6, 97.0)	0.003
Specificity, %	97.6 (93.9, 99.3)	97.0 (93.0, 99.0)	97.6 (93.9, 99.3)	1.000
Primary malignancy except HCC
AUC	0.899 (0.793, 1.000)	0.892 (0.783, 1.000)			0.844
Sensitivity, %	53.3 (26.6, 78.7)	46.7 (21.3, 73.4)	60.0 (32.3, 83.7)	0.688
Specificity, %	97.9 (94.6, 99.4)	96.8 (93.1, 98.8)	96.2 (92.4, 98.5)	0.727

^aP-value was calculated between Model A (seven-classification) vs. Reader consensus using the McNemar's test.