Table 2 Networks performance on Colombian test dataset with lung segmentation.
From: Cov-caldas: A new COVID-19 chest X-Ray dataset from state of Caldas-Colombia
CNN architecture | Accuracy | Specificity | F1 score | Recall | Precision | FN | FP | TN | TP |
|---|---|---|---|---|---|---|---|---|---|
Seed 1 | |||||||||
InceptionV3 | 84.88 | 82.50 | 86.68 | 86.68 | 86.68 | 53 | 53 | 250 | 345 |
VGG16 | 84.59 | 80.20 | 86.63 | 87.94 | 85.37 | 48 | 60 | 243 | 350 |
DenseNet201 | 84.31 | 77.56 | 86.62 | 89.45 | 83.96 | 42 | 68 | 235 | 356 |
Seed 2 | |||||||||
InceptionResNetV2 | 84.96 | 83.28 | 86.61 | 86.27 | 86.95 | 53 | 50 | 249 | 333 |
DenseNet201 | 84.09 | 81.27 | 85.94 | 86.27 | 85.60 | 53 | 56 | 243 | 333 |
InceptionV3 | 83.36 | 84.28 | 84.84 | 82.64 | 87.16 | 67 | 47 | 252 | 319 |
Seed 3 | |||||||||
VGG16 | 83.70 | 79.00 | 86.27 | 86.98 | 85.58 | 56 | 63 | 237 | 374 |
InceptionResNetV2 | 82.60 | 80.67 | 85.04 | 83.95 | 86.16 | 69 | 58 | 242 | 361 |
InceptionV3 | 81.10 | 86.67 | 82.79 | 77.21 | 89.25 | 98 | 40 | 260 | 332 |
