Table 7.

Three-class classification results: Dataset 6

CNN	Accuracy (%)	Sensitivity (%)	Specificity (%)	Precision (%)	$F_1$ score	AUC
90% training and 10% testing
AlexNet	97.59 ± 0.60	95.45 ± 4.55	97.76 ± 0.37	98.55 ± 2.51	0.969 ± 0.014	0.998 ± 0.004
GoogLeNet	96.09 ± 2.30	96.97 ± 2.62	96.02 ± 2.28	98.48 ± 2.62	0.977 ± 0.023	0.999 ± 0.004
SqueezeNet	97.47 ± 1.31	98.48 ± 2.62	97.39 ± 1.30	94.20 ± 2.51	0.963 ± 0.026	0.999 ± 0.009
50% training and 50% testing
AlexNet	95.89 ± 0.42	92.11 ± 6.31	96.20 ± 0.66	96.66 ± 2.30	0.942 ± 0.029	0.998 ± 0.000
GoogLeNet	96.07 ± 0.63	95.23 ± 1.51	96.14 ± 0.72	95.24 ± 1.72	0.952 ± 0.015	0.999 ± 0.001
SqueezeNet	95.95 ± 0.52	92.11 ± 3.02	96.26 ± 0.63	96.75 ± 1.20	0.943 ± 0.014	0.998 ± 0.001