Table 6.

Application of AI in colposcopy.

Reference	Year	Aim of the study	Number of subjects	Methods	Images	Results
Kim E et al. (68)	2013	Detection of CIN2+	2000images	SVM	Cervicography	Sensitivity 75.00%Specificity 75.00%
Song et al. (69)	2015	Detection of CIN2+	7669patients	MCNN	Cervicography	Accuracy 80.00%Sensitivity 83.21%Specificity 94.79%
Hu et al. (70)	2019	Detection of CIN2+	9406patients	Faster-CNN	Cervicography	AUC 0.91
Chao et al. (71)	2020	Detection lesions need to biopsy and classification	791 patients	CNN	Optical colposcopy image	Sensitivity 85.20%Specificity 88.20%AUC 0.947
Asiedu et al. (72)	2019	Classification of cervical lesions	134 patients	SVM	Digital colposcopy images	Accuracy 80.00%Sensitivity 81.30%Specificity 78.60%
Yuan et al. (26)	2020	Classification of cervical lesions	22330images	CNN	Digital colposcopy images	Sensitivity 85.38%Specificity 82.62%
Miyagi et al. (73)	2019	Classification of cervical lesions	253patients	CNN	Traditional colposcopy images	Accuracy 83.30%Sensitivity 95.60%
Miyagi et al. (23)	2019	Classification of cervical lesions	310images	CNN	Traditional colposcopy images	Accuracy 82.30%Sensitivity 80.00%Specificity 88.20%
Xue et al. (74)	2020	Classification of cervical lesions	19435patients	CAIADS	Digital colposcopy images	LSIL Sensitivity 90.50%Specificity 51.80%HSIL Sensitivity 71.90%Specificity 93.90%
Yue et al. (75)	2020	Classification of cervical lesions	4753images	CNN,	cervigram images	Accuracy 96.13%Sensitivity 95.09%Specificity 98.22%AUC 0.94
Venkatesan et al. (76)	2021	Classification of cervical lesions	5679images	CNN	colposcopy photographs	Accuracy 83.30%Sensitivity 95.60%
Peng et al. (77)	2021	Classification of cervical lesions	300images	VGG16	colposcopy images	Accuracy 86.30%Sensitivity 84.10%Specificity 89.80%

CAIADS, Colposcopic Artificial Intelligence Auxiliary Diagnostic System.