Table 3.

Performance results of pre-trained CNN models for analysis of the ISIC 2019 dataset image for early diagnosis and differentiation of skin cancer and other skin lesions.

Models	Type of Lesion	AUC %	Sensitivity %	Accuracy %	Precision %	Specificity %
AlexNet	Scc	50.2	45.4	45.2	54.8	98.5
	Akiec	51.5	46.3	44.8	57.4	99.2
	Bcc	89.3	86.3	85.6	84.3	97.6
	Bkl	88.1	85.9	86.3	84.7	97.9
	Df	84.9	72.8	72.9	68.6	99.6
	Mel	91.2	89.2	89.5	85.9	97.4
	Nv	94.8	95.2	95.3	95.2	94.8
	Vasc	78.6	65.2	64.7	68.8	99.6
GoogLeNet	Scc	46.5	39.3	38.9	49	99.1
	Akiec	88.2	87.9	48.3	57.9	98.4
	Bcc	90.4	89.2	89	84	96.9
	Bkl	87.6	86.4	86.3	82.8	97.5
	Df	68.2	58.2	58.3	66.7	99.5
	Mel	89.3	87.7	87.7	89	98.1
	Nv	97.1	95.8	95.7	94.9	95.4
	Vasc	72.5	65.3	64.7	78.7	99.6
VGG16	Scc	51.5	44.2	43.7	55	99.2
	Akiec	52.3	44.3	43.7	56.7	98.7
	Bcc	89.6	87.9	87.8	83.2	97.3
	Bkl	88.3	86.2	85.9	85.6	97.9
	Df	59.8	46.4	45.8	50	99.6
	Mel	92.9	88.7	88.6	86.4	96.7
	Nv	96.6	94.6	95.3	94.6	93.8
	Vasc	77.2	65.4	64.7	80.5	98.6