TABLE 1.

The performances of the models of the models.

Models	Train/Test	Accuracy	AUC	95% CI	Sensitivity	Specificity	F1-score
Ensemble DL	Train	94.20%	0.978	(0.958–0.998)	90.79%	98.39%	94.52%
	Test	94.12%	0.980	(0.968–1.000)	89.47%	100.00%	94.44%
Alexnet	Train	85.40%	0.900	(0.847–0.953)	90.79%	78.69%	87.34%
	Test	85.71%	0.878	(0.754–1.000)	89.47%	81.25%	87.18%
Googlenet	Train	84.67%	0.912	(0.867–0.958)	77.63%	93.44%	84.89%
	Test	80.00%	0.872	(0.759–0.985)	68.42%	93.75%	78.79%
Resnet18	Train	80.29%	0.835	(0.765–0.906)	82.89%	77.05%	82.35%
	Test	77.14%	0.816	(0.673–0.959)	63.16%	93.75%	75.00%
Vgg11	Train	84.67%	0.920	(0.878–0.963)	85.52%	83.61%	86.10%
	Test	85.71%	0.921	(0.836–1.000)	94.74%	75.00%	87.80%