Table 2.

Technical aspects of the included studies.

Study	Year	Type of internal validation	DLa model	Optimal DL algorithms	Internal validation or test sets				External validation or test sets
					TPb	FPc	FNd	TNe	TP	FP	FN	TN
Wang et al [18]	2020	10-fold cross-validation	GoogLeNet Inception version 3	CNNf	8	0	2	60	NRg	NR	NR	NR
Kharazmi et al [19]	2018	Random split test set	SAEh with softmax classifier	Autoencoder	128	18	22	282	NR	NR	NR	NR
Maurya et al [20]	2024	Random split test set	Hybrid model of EfficientNet-B5+random forest classifier	CNN	190	4	5	196	NR	NR	NR	NR
UdriȘtoiu et al [21]	2020	Random split test set	4-layer CNN	CNN	185	11	2	1051	NR	NR	NR	NR
Zhu et al [22]	2021	NR	Modified CNN model based on EfficientNet-B4	CNN	473	275	14	12,841	22	1	3	174
Serrano et al [23]	2022	Random split test set	VGG16 with custom fully connected layers	CNN	302	15	2	245	NR	NR	NR	NR
Cheng et al [24]	2011	Leave-one-out cross-validation	Standard backpropagation neural network	MLPi	55	22	4	130	NR	NR	NR	NR
Maurya et al [25]	2024	Random split test set	Hybrid model of EfficientNet-B5+U-Net+random forest classifier	CNN	191	7	4	193	NR	NR	NR	NR
Radhika and Chandana [26]	2023	Random split test set	MSCD-Netj model	CNN	550	95	72	4349	NR	NR	NR	NR
Maron et al [27]	2019	Random split test set	ResNet-50	CNN	52	4	8	236	NR	NR	NR	NR
Naeem et al [28]	2022	Leave-one-out cross-validation	SCDNetk (VGG16+CNN)	CNN	550	95	72	4349	NR	NR	NR	NR
Ali et al [29]	2023	10-fold cross-validation	26-layer CNN model	CNN	197	12	3	188	NR	NR	NR	NR
Panthakkan et al [30]	2022	5-fold cross-validation	Concatenated Xception–ResNet-50 model	CNN	323	8	7	1762	NR	NR	NR	NR
Priyeshkumar et al [31]	2024	Random split test set	Mg-EDCFl model	DCFm	296	24	4	1776	NR	NR	NR	NR
Minagawa et al [32]n	2020	Random split test set	Inception–ResNet version 2	CNN	9	1	3	37	NR	NR	NR	NR
Minagawa et al [32]o	2020	Random split test set	Inception–ResNet version 2	CNN	12	0	0	38	NR	NR	NR	NR

^aDL: deep learning.

^bTP: true positive.

^cFP: false positive.

^dFN: false negative.

^eTN: true negative.

^fCNN: convolutional neural network.

^gNR: not reported.

^hSAE: sparse autoencoder.

ⁱMLP: multilayer perceptron.

^jMSCD-Net: multiclass skin cancer detection network.

^kSCDNet: skin cancer detection classifier network.

^lMg-EDCF: multigrained enhanced deep cascaded forest.

^mDCF: deep cascaded forest.

ⁿShinshu test set.

^oInternational Skin Imaging Collaboration test set.