Table 3.

The application of AI in binary classification for skin lesions.

Authors	Reference	Year	Country	AI Algorithm Model	The Purpose of AI Algorithm	Image (Datasets) Recourse	No. of Images in Datasets	Types of Images	Accuracy/ Precision (%)	Sensitivity/Recall (%)	Specificity (%)
Filho et al.	[51]	2018	Germany	Structural Co-occurrence matrix	Classification of melanoma	ISIC-2016, 2017, PH2	3100	Dermoscopy	89.93–99	89.9–99.2	95.15–99.4
Marchetti et al.	[89]	2018	USA	Non-learned approaches and machine learning methods	Classification of melanoma and comparison of the performance of AI model with 8 dermatologists	ISIC-2016	1279	Dermoscopy	85–86	46–70	88–92
Roffman et al.	[94]	2018	USA	Artificial neural network	Detection of non-melanoma skin cancer	NHIS 1997–2015	462,630	Macroscopy	81	86.2–88.5	62.2–62.7
Alzubaidi	[95]	2021	Australia	Transfer learning model	Discrimination of skin cancer and normal skin	ISIC-2016–2020, Med- Node, Dermofit	>200,000	Dermoscopy	89.69–98.57	85.60–97.90	N/A
Guimarães et al.	[96]	2020	Germany	Convolutional neural networks	Diagnosis of atopic dermatitis	Multiphoton tomography Images	3663	Multiphoton tomograph	97.0 ± 0.2	96.6 ± 0.2	97.7 ± 0.3
Ho. et al.	[97]	2020	USA	Deep neural network	Image segmentation of plexiform neurofibromas	MRI images	35	MRI	N/A	N/A	N/A
Fink et al.	[98]	2018	Germany	Edge-preserving thresholding automated shape recognition	Classification of psoriasis and measurement of lesion area andseverity index	Clinical images	10 patients	Macroscopy	N/A	N/A	N/A
Fink et al.	[99]	2019	Germany	Edge-preserving thresholding automated shape recognition	Validation of the precision and reproducibility of algorithm in PASI measurements	Clinical images	120 patients	Macroscopy	N/A	N/A	N/A
Schnuerle et al.	[100]	2017	Switzerland	Support vector machines	Detection of hand eczema	Clinical images	N/A	Macroscopy	74.5–89.29	48–71.43	77.24–93.63
Gao et al.	[101]	2020	Chinas	Deep learning network architecture (ResNet-50)	Detection for fungal skin lesion	Clinical images	292	Macroscopy	N/A	95.2–99.5	91.4–100
Bashat et al.	[102]	2018	Israel	N/A	Differentiation of benign and malignant neurofibroma	MRI images	30	MRI	80	72	87
Duarte et al.	[103]	2014	Portugal	Support vector machines	Classification of whole-brain grey and white matter of MRI between NF1 patients and normal person	T1-weighted MRI scans	99	MRI Images	94	92	96
Meienberger et al.	[104]	2019	Switzerland	Convolutional neural networks (Net 16)	Establishment of an accurate and objective psoriasis assessment method	Clinical images	203	Macroscopy	92	N/A	N/A
Gustafson et al.	[105]	2017	USA	Electronic health record based phenotype algorithm	Identification of atopic dermatitis and comparison of the performance of AI model with 4 dermatologists	Clinical images	562	N/A	N/A	53.6–75	N/A
Luo et al.	[106]	2020	China	Cycle-consistent adversarial networks	Classification of vitiligo skin lesion	Clinical Images	80,000	Macroscopy	85.69	80.73	66.2
Makena et al.	[107]	2019	USA	Convolutional neural networks	Segmentation of vitiligo skin lesion	RGB images of vitiligo lesions	308	Macroscopy (UV/natural light)	74–88.7	N/A	N/A