Table 7.

Performance results of CNN-ANN hybrid models based on the fused CNN models for early diagnosis and differentiation of skin cancer and other skin lesions.

Fusion Features	Classifier	Type of Lesion	AUC %	Sensitivity %	Accuracy %	Precision %	Specificity %
AlexNet-GoogLeNet	ANN	Scc	87.8	84.4	84.1	84.1	99.5
		Akiec	85.6	83.9	84.5	89.1	99.6
		Bcc	97.6	97.3	97.1	96.3	98.7
		Bkl	98.2	96.8	96.6	95.1	98.5
		Df	79.6	56.4	56.3	71.1	100
		Mel	94.2	92.9	93.3	91.7	97.8
		Nv	95.4	97.4	97.2	97.2	97.4
		Vasc	88.2	69.3	68.6	83.3	99.6
GoogLeNet-VGG16	ANN	Scc	87.2	86.4	85.7	83.7	99.5
		Akiec	82.4	78.3	78.2	87.7	99.7
		Bcc	95.7	96.2	96.4	93	98.8
		Bkl	96.2	94.4	93.9	96	98.7
		Df	82.1	71.3	70.8	70.8	100
		Mel	96.1	93.8	94.4	91.8	97.8
		Nv	97.8	96.9	96.7	97.8	98.4
		Vasc	87.9	78.4	78.4	74.1	99.7
AlexNet-VGG16	ANN	Scc	92.1	88.2	88.1	86	99.8
		Akiec	93.1	87.8	87.9	89.5	100
		Bcc	97.6	98.1	97.9	97.2	99.5
		Bkl	98.9	98.4	97.5	95.7	98.8
		Df	89.4	71.3	70.8	79.1	99.7
		Mel	95.2	93.2	92.9	90.2	97.6
		Nv	97.1	96.4	96.5	97.6	98.4
		Vasc	91.6	74.9	74.5	84.4	100
AlexNet-GoogLeNet-VGG16	ANN	Scc	91.9	86.2	85.7	83.7	99.6
		Akiec	94.5	87.1	86.8	88.3	100
		Bcc	96.8	96.3	96.2	97	100
		Bkl	97.4	95.8	96.4	95.5	98.7
		Df	88.3	73.4	72.9	74.5	99.5
		Mel	97.8	96.2	96.3	93.2	98.3
		Nv	98.2	98.1	97.8	98.9	99.4
		Vasc	90.4	78.1	78.4	78.4	100