Table 5.
Binary classification accuracy (%), number of features, and other performance metrics before and after feature selection for ResNet-50 and Inception-ResNet CNNs.
| Method | No of features | Accuracy (%) | Sensitivity | Specificity | Precision | F1-score | MCC |
|---|---|---|---|---|---|---|---|
| ResNet-50 | |||||||
| DWT | 1,074 | 100 | 1 | 1 | 1 | 1 | 1 |
| Relief-F | 578 | 100 | 1 | 1 | 1 | 1 | 1 |
| IG | 578 | 100 | 1 | 1 | 1 | 1 | 1 |
| Inception-ResNet | |||||||
| DWT | 818 | 100 | 1 | 1 | 1 | 1 | 1 |
| Relief-F | 200 | 100 | 1 | 1 | 1 | 1 | 1 |
| IG | 550 | 100 | 1 | 1 | 1 | 1 | 1 |