Table 3.
The test results of RMT-Net compared to the other four models.
| Datasets | Methods | Size(M) | Specificity (%) | Sensitivity (%) | Test _ acc (%) | Speed (ms) |
|---|---|---|---|---|---|---|
| Four classes (X-ray images) | ResNet-50 | 285 | 97.24 | 92.84 | 93.14 | 12.24 |
| VGGNet-16 | 146 | 93.54 | 92.25 | 92.62 | 10.09 | |
| i-CapsNet | 84 | 92.62 | 92.86 | 93.15 | 8.58 | |
| MGMADS-3 | 43.6 | 98.06 | 96.60 | 96.75 | 6.09 | |
| RMT-Net | 40.8 | 98.26 | 98.08 | 97.65 | 5.46 | |
| Binary classes (CT images) | ResNet-50 | 275 | 96.14 | 95.48 | 95.25 | 10.37 |
| VGGNet-16 | 154 | 96.45 | 94.16 | 94.38 | 7.83 | |
| i-CapsNet | 82 | 94.67 | 95.32 | 95.62 | 5.79 | |
| MGMADS-3 | 43.6 | 98.17 | 98.05 | 98.25 | 4.23 | |
| RMT-Net | 38.5 | 99.34 | 98.76 | 99.12 | 4.12 |
Bold value highlights the gain effect of our method in the table.