Table 2.
DL applications on detecting COVID-19 from chest X-ray images
| Literature | Data set without augmentation | DL tool | Accuracy | Specificity | Sensitivity | F1 score |
|---|---|---|---|---|---|---|
| Loey et al. [16] |
3 Classes N/C/P 69/79/79 |
ALexnet | 85.19 | - | 85.19 | 85.19 |
| Googlenet | 81.48 | - | 81.48 | 81.46 | ||
| Resnet18 | 81.48 | - | 81.48 | 84.66 | ||
| Apostolopoulos and Mpesiana [13] |
3 classes Dataset_1 N/C/BP 504/224/700 |
VGG19 | 93.48 | 98.75 | 92.85 | |
| MobileNet (v2) | 92.85 | 97.09 | 99.10 | |||
| Inception | 92.85 | 99.70* | 12.94* | |||
| Xception | 92.85 | 99.99* | 0.08* | |||
| Inception ResNet v2 | 92.85 | 99.83* | 0.01* | |||
|
3 classes Dataset_2 N/C/BP+VP 504/224/714 |
MobileNet (v2) | 94.72 | 96.46 | 98.66 | - | |
| Ucar and Korkmaz [6] |
3 classes N/C/P 1583/76/4290 |
SqueezeNet with raw dataset | 76.37 | 79.93 | - | 98.25 |
|
3 classes N/C/P 1536/1536/1536 |
SqueezeNet with augmented dataset | 98.26 | 99.13 | - | 98.25 | |
| Ozturk et al. [17] |
3 Classes N/C/P 500/127/500 |
Darknet | 87.02 | 92.18 | 85.35 | 87 |
| Proposed study |
3 Classes N/C/VP 350/210/350 |
COV19-CNNet | 94.28 | 96.94 | 94.33 | 94.20 |
| COV19-ResNet | 97.61 | 98.72 | 97.61 | 97.62 | ||
|
N: Normal (healthy) C: COVID-19 NC: Non-COVID-19 |
P: Pneumonia VP: Viral pneumonia BP: Bacterial pneumonia |
|||||