TABLE III. Comparison of Performances With Other State-of-The-Art Networks on COVID Lesion Segmentation on Dataset-2.
| Network | Sensitivity (%) | Specificity (%) | Dice (%) | IoU (%) |
|---|---|---|---|---|
| Unet [16] | 75.9 0.34 |
88.9 0.12 |
79.3 0.26 |
74.9 0.18 |
| MultiResUnet [30] | 77.2 0.33 |
90.3 0.24 |
82.7 0.28 |
77.4 0.15 |
| Attention-Unet [31] | 81.1 0.29 |
92.2 0.11 |
85.1 0.14 |
79.6 0.28 |
| CPF-Net [29] | 78.9 0.27 |
91.7 0.14 |
84.4 0.25 |
79.3 0.25 |
| Gated-Unet [32] | 81.4 0.24 |
92.5 0.19 |
85.6 0.19 |
80.2 0.16 |
| Inf-Net [12] | 82.7 0.26 |
94.8 0.21 |
86.9 0.34 |
81.1 0.18 |
| TA-SegNet (ours) | 88.5 0.22 |
98.9 0.14 |
90.2 0.17 |
86.4 0.19 |