Table 3.
Some of experimental results for COVID-19 diagnosis
| Task | Author | Network | Dataset | Interpretability method | Accuracy | Precision | Sensitivity/recall | Specificity | F1 score |
|---|---|---|---|---|---|---|---|---|---|
| COVID-19 Detection | Alshazly et al. [79] | ResNet101 | SARS-CoV-2 CT-scan |
t-SNE, Grad-CAM |
99.40% | 99.60% | 99.80% | 99.60% | 99.40% |
| DenseNet201 | COVID19-CT | 92.90% | 91.30% | 93.70% | 92.20% | 92.50% | |||
| COVID-19 Detection | Shi et al. [74] | - |
COVID-19 X-Ray dataset (private) |
Attention | 0.9411 | 0.9673 | 0.978 | – | 0.9726 |
|
COVID-19 CT dataset (private) |
0.8654 | 0.913 | 0.8513 | – | 0.8811 | ||||
| COVID-19 Detection | Brunese et al. [76] | VGG-16 | Private dataset | Grad-CAM | 0.98 | – | 0.87 | 0.94 | 0.89 |
| COVID-19 Detection | Karim et al. [78] | Ensemble model |
Private chest X-ray dataset (balanced /imbalanced dataset) |
LRP, Grad-CAM++ |
– |
0.904 0.877 |
0.905 0.881 |
0.905 0.879 |
– |
| COVID-19 Classification | Wu et al. [75] | Res2Net | Private COVID-CS dataset | Activation mapping | – | – | 95% | 93% | – |