Table 2.
Comparison of studies with similar convolutional neural network architecture.
| Study | Objectives | Dataset | F1 score |
|---|---|---|---|
| Xu et al. (2017) | Segmentation of colon glands | GLAS challenge (165 images) | 0.893–0.843 |
| Xu et al. (2016) | Nuclei segmentation | 537 images from Case Western Reserve University | 0.858–0.771 |
| Korbar et al. (2017) | Deep Neural Network Visualization to Interpret WSI Analysis Outcomes for Colorectal Polyps | 176 WSIs from Dartmouth-Hitchcock Medical Center | 0.925–0.841 |
| MIMO—Net35 | Various studies | Various studies | 0.913–0.724 |
| DeepLab v3+36 | Various studies | Various studies | 0.862–0.764 |
| SegNet37 | Various studies | Various studies | 0.858–0.783 |
| FCN—837 | Various studies | Various studies | 0.783–0.692 |
| Qritive Colon AI (current study) | Glandular segmentation deep learning model to detect high risk colorectal polyps | WSIs produced from 294 colorectal specimens from Singapore General Hospital | 0.974–0.856 |
GLAS Gland Segmentation in Colon Histology Images Challenge Contest, WSI whole slide images.