Table 5.

Results obtained after applying the proposed scheme.

Architecture	Filter	Accuracy	Precision	Recall	Specificity	F1-Score	MCC
AlexNet	Sobel	79.50	61.96	90.48	74.45	73.55	60.51
	LoG	82.00	69.57	88.89	78.12	78.05	64.54
	Gabor	84.50	68.48	96.92	78.52	80.25	70.90
GoogleNet	Sobel	75.00	56.52	83.87	71.01	67.53	50.93
	LoG	76.50	54.35	90.91	71.03	68.03	55.50
	Gabor	80.50	60.87	94.92	74.47	74.17	63.49
VGG-19	Sobel	82.50	66.30	93.85	77.04	77.71	66.61
	LoG	84.00	69.57	94.12	78.79	80.00	69.29
	Gabor	86.00	71.74	97.06	80.30	82.50	73.53
ShuffleNet	Sobel	95.00	91.30	97.67	92.98	94.38	90.05
	LoG	96.50	93.48	98.85	94.69	96.09	93.05
	Gabor	98.00	98.91	96.81	99.06	97.85	96.00
DenseNet-121	Sobel	96.50	94.57	97.75	95.50	96.13	92.98
	LoG	96.00	93.48	97.73	94.64	95.56	92.00
	Gabor	97.50	98.91	95.79	99.05	97.33	95.02
DenseNet-201	Sobel	97.00	94.57	98.86	95.54	96.67	94.02
	LoG	97.50	95.65	98.88	96.40	97.24	95.00
	Gabor	98.50	98.91	97.85	99.07	98.38	96.99

* The bold numbers indicate the maximum values in the considered architectures.