Table 5.
Application of AI in cytology to detect CIN.
| Reference | Year | N | Methods | Databases | Results |
|---|---|---|---|---|---|
| Yu et al. (32) | 2018 | 1839 | Risk score algorithm | Cytological image HPV testing | CIN2+ AUC 0.710CIN3+ AUC 0.740 |
| Bao et al. (60) | 2020 | 703103 | DL | Cytological image | CIN1+ Sensitivity 88.9%Specificity 95.8%CIN2+ Sensitivity 90.10%Specificity 94.80%CIN3+Sensitivity 90.90%Specificity 94.40% |
| Bao et al. (61) | 2020 | 2145 | ResNet | Cytological image | CIN2+ AUC 0.762CIN3+ AUC 0.755 |
| Wang et al. (62) | 2020 | 143 | DL | whole slide images (WSIs) | precision 93.00%recall 90.00%, F-measure 88.00% |
| Holmström O et al. (59) | 2021 | 740 | DL | Cytological image | HSIL+ AUC 0.970Sensitivity 85.7%Specificity 98.5% |
| Zhu et al. (63) | 2021 | 980 | AIATBS | Cytological imageBiopsy diagnosis results | Sensitivity 94.74% |
DL, deep learning; CNN, convolutional neural network; AIATBS, artificial intelligence-assisted TBS; CIN, cervical intraepithelial neoplasia; AUC, area under the curve; HSIL, high squamous intraepithelial lesion.