Table 4.

Fusion models performance.

Fusion Models	AUC (95%CI)	Accuracy	Sensitivity	Specificity	Precision	Recall	F1-score
Vgg19
Training	0.999 (0.999–1.000)	0.985	0.993	0.975	0.981	0.993	0.987
Validation	0.972 (0.946–0.997)	0.935	0.935	0.934	0.947	0.935	0.941
resnet50
Training	0.995 (0.988–1.000)	0.978	0.987	0.967	0.974	0.987	0.981
Validation	0.923 (0.875–0.971)	0.891	0.922	0.852	0.888	0.922	0.904
GoogLeNet
Training	0.997 (0.995–1.000)	0.987	0.987	0.988	0.990	0.987	0.989
Validation	0.947 (0.914–0.980)	0.861	0.908	0.803	0.852	0.908	0.879
Inception-v3
Training	0.988 (0.982–0.994)	0.942	0.961	0.918	0.936	0.961	0.948
Validation	0.883 (0.826–0.940)	0.790	0.883	0.672	0.773	0.883	0.824

AUC, area under the curve; 95%CI, 95 % confidential interval.