Table 3. SARS-CoV-2 CT Scan dataset average results for the baseline by Silva, et al. (19), end-to-end, and embedding classification approaches. The highest values for each metric across all experiments are highlighted in bold.
| Method | Configuration | Acc (%) | Pre (%) | Sens (%) | F1 (%) | AUC (%) |
|---|---|---|---|---|---|---|
| Silva, et al. 19 | EfficientNetB0 | 86.6 ± 10.1 | 79.7 ± 20.9 | 94.8 ± 4.50 | - | - |
| End-to-end | VGG16 | 92.33 ± 4.81 | 89.70 ± 6.74 | 88.96 ± 6.57 | 89.89 ± 6.38 | 98.20 |
| ResNet-152 | 86.05 ± 1.43 | 85.52 ± 1.33 | 76.02 ± 4.01 | 79.01 ± 3.37 | 88.51 | |
| Embedding | ResNet-152 + RF | 90.70 ± 2.80 | 91.38 ± 2.83 | 95.62 ± 2.85 | 93.42 ± 2.38 | 88.82 |
| ResNet-152 + SVM | 91.40 ± 2.48 | 95.77 ± 2.83 | 91.58 ± 2.41 | 93.63 ± 2.80 | 91.28 |