Table 1.
Comparison of our proposed method with existing state-of-the-art methods using COVIDx dataset, the best performances are indicated in bold.
| Method | Accuracy (%) | Sensitivity (%) | Specificity (%) | Precision (%) |
|---|---|---|---|---|
| Zhong et al.66 | 86.94 | 76.38 | 95.83 | 78.01 |
| Deep-COVID18 | 82.83 | 71.18 | 93.95 | 69.49 |
| CoroNet68 | 89.02 | 90.62 | 95.36 | 79.09 |
| EDL-COVID67 | 95.0 | 96.0 | 95.35 | 94.1 |
| PbCNN17 | 91.9 | 92.5 | 96.4 | 76.9 |
| Ismael et al.19 | 91.3 | 95.0 | 94.0 | 88.8 |
| COVID-Net.9 | 93.3 | 91.0 | 99.4 | 98.9 |
| Brunese et al.69 | 84.6 | 67.0 | 95.5 | 88.1 |
| nCOVnet70 | 87.3 | 82.0 | 96.0 | 91.1 |
| DarkCovidNet40 | 88.6 | 89.0 | 97.5 | 94.6 |
| Rajaraman et al.71 | 82.6 | 64.0 | 96.0 | 88.8 |
| DAM-Net | 97.22 | 96.87 | 99.12 | 95.54 |