Table 2.
Comparison of CNN–RNN models for multi-classification network.
| Model (CNN + RNN + Enhancement) |
Patient Status | ACC (%) |
Precision (%) |
Recall (%) |
F1-Score (%) |
Training Times |
Predict Times (/Image) |
|---|---|---|---|---|---|---|---|
| ResNet152V2 + GRU + Original |
COVID-19 Pneumonia Normal Overall |
94.14 98.95 93.65 93.37 |
90.58 98.11 92.49 93.73 |
94.64 98.31 87.36 93.44 |
92.57 98.21 89.85 93.54 |
99 m 57 s |
0.21 s |
| VGG19 + LSTM + Normalization |
COVID-19 Pneumonia Normal Overall |
92.75 99.35 92.47 92.29 |
89.68 99.26 89.14 92.60 |
91.76 98.52 87.26 92.69 |
90.71 98.89 88.19 92.51 |
115 m 1 s |
0.16 s |
| DenseNet121 + LSTM + Normalization |
COVID-19 Pneumonia Normal Overall |
91.45 99.23 90.86 90.77 |
84.43 98.22 92.89 91.85 |
95.44 99.16 77.59 90.73 |
89.60 98.69 84.55 90.95 |
103 m 55 s |
0.08 s |