Table 6.

Performance results of CNN-RF hybrid models based on ROI using the GAC method for early diagnosis and differentiation of skin cancer and other skin lesions.

Hybrid-Models	Type of Lesion	AUC %	Sensitivity %	Accuracy %	Precision %	Specificity %
AlexNet-RF	Scc	89.2	82.3	81.7	79.8	99.1
	Akiec	84.6	80.4	79.9	83.7	99.3
	Bcc	96.7	95.7	96.1	94.4	89.7
	Bkl	95.3	93.8	93.5	95.5	98.5
	Df	87.1	73.4	72.9	64.8	99.6
	Mel	93.1	92.9	93.4	90.9	97.8
	Nv	94.5	96.6	96.5	97.6	98.4
	Vasc	81.9	78.4	78.4	76.9	99.7
GoogLeNet-RF	Scc	90.2	83.4	82.5	75.4	98.7
	Akiec	82.1	79.3	79.3	81.2	99.2
	Bcc	94.3	94.3	94	95.6	98.6
	Bkl	91.2	94.8	94.7	94.3	99.1
	Df	79.6	33.4	33.3	59.3	99.8
	Mel	94.2	96.4	96.2	92.1	97.5
	Nv	96.8	97.8	98.1	98.7	99.5
	Vasc	85.4	69.3	68.6	74.5	100
VGG16-RF	Scc	82.5	79.3	79.4	76.9	99.2
	Akiec	86.2	75.2	75.3	79.9	98.8
	Bcc	96.1	94.3	94.4	93.3	98.6
	Bkl	94.5	95.4	94.9	94.9	98.5
	Df	67.2	31.4	31.3	50	100
	Mel	92.9	94.3	94.2	88.8	97.2
	Nv	97.2	96.9	97.4	98.9	99.3
	Vasc	87.2	76.4	76.5	79.6	99.7