Table 3.

Comparison of discriminant performance between 3D-ResNet and 3 radiologists in the test set

Performance	3D-ResNet	Reader1-10y	Reader2-5y	Reader3-3y
AUC	0.86	0.69	0.65	0.63
Accuracy	0.83	0.80	0.77	0.75
Specificity	0.57	0.46	0.39	0.39
Sensitivity	0.92	0.92	0.91	0.87
NTM-LD_Precision	0.73	0.68	0.61	0.52
NTM-LD_F1 score	0.64	0.55	0.48	0.45
MTB-LD_Precision	0.86	0.83	0.81	0.80
MTB-LD_F1 score	0.89	0.87	0.86	0.83
sec/CT	0.04 ~ 0.06	58	62	71