Research Article
Radiologists versus Deep Convolutional Neural Networks: A Comparative Study for Diagnosing COVID-19
Table 3
Performance metrics of the employed deep neural network models and the radiologists on the testing images.
| | Accuracy | Sensitivity | Specificity | Precision | F1-score |
| | Rad 1 | 76.4% | 79.1% | 80.3% | 80.4% | 79.74% | | Rad 2 | 75.9% | 74.1% | 78.7% | 79.2% | 76.56% | | ResNet-18 | 91.3% | 90.1% | 94.3% | 91.4% | 90.74% | | ResNet-50 | 97.2% | 93.1% | 94.3% | 96.1% | 94.58% | | DenseNet-201 | 97.8% | 98.1% | 97.3% | 98.4% | 98.25% |
|
|
