Table 1.
Representative deep learning-based methods for Ct-scan COVID-19 diagnosis.
| Reference | Methodology | Performance metrics | Task |
|---|---|---|---|
| Keno et al. [3] | Used transfer learning to train 3D-U-Net model using an 18-layer 3D ResNet | Dice 0.679 | Segmentation |
| Adnan et al. [4] | Used SegNet and U-Net deep learning networks | PixAcc. 0.908 U-Net 0.907 SegNet | Segmentation |
| Chen et al. [5] | Used UNet++ to train in Keras in an image-to-image manner | PixAcc. 0.95 | Detection |
| Li et al. [6] | Introduced COVNet | PixAcc. 0.96 | Detection |
| Ying et al. [7] | Proposed DRE-Net | PixAcc. 0.86 | Detection |
| Garain et al. [8] | Proposed SNNs | PixAcc. 0.91 | Detection |
| Stefano et al. [9] | Proposed C-ENET | Dice. 0.75 | Segmentation |
| Zhang et al. [10] | Improved DCNN model | PixAcc. 0.93 | Detection |
| Shouliang et al. [11] | Proposed DR-MIL method | PixAcc. 0.95 | Detection |