Table 4.
Performance of the FCONet frameworks based on the four pretrained models on the testing data set.
| Model and data group | Sensitivity, % | Specificity, % | Accuracy, % | AUCa | P value | ||
| ResNet-50 | <.001 | ||||||
|
|
COVID-19b pneumonia | 99.58 | 100.00 | 99.87 | 1.00 |
|
|
|
|
Other pneumonia | 97.42 | 99.81 | 99.00 | 0.99 |
|
|
|
|
Nonpneumonia | 100.00 | 98.63 | 99.12 | 0.99 |
|
|
| VGG16 | <.001 | ||||||
|
|
COVID-19 pneumonia | 100.00 | 99.64 | 99.75 | 1.00 |
|
|
|
|
Other pneumonia | 100.00 | 99.81 | 99.87 | 0.99 |
|
|
|
|
Nonpneumonia | 100.00 | 99.80 | 99.87 | 0.99 |
|
|
| Xception | <.001 | ||||||
|
|
COVID-19 pneumonia | 97.91 | 99.29 | 98.87 | 0.99 |
|
|
|
|
Other pneumonia | 98.52 | 99.05 | 98.87 | 0.99 |
|
|
|
|
Nonpneumonia | 100.00 | 100.00 | 100.00 | 1.00 |
|
|
| Inception-v3 | <.001 | ||||||
|
|
COVID-19 pneumonia | 88.28 | 97.68 | 94.87 | 0.97 |
|
|
|
|
Other pneumonia | 94.10 | 95.83 | 95.24 | 0.98 |
|
|
|
|
Nonpneumonia | 98.27 | 97.25 | 97.62 | 0.99 |
|
|
aAUC: area under the curve.
bCOVID-19: coronavirus disease.