Table 8 Grade 3 severity detection using performance metrics of various DL methods.
Algorithms | Performance metrics | |||||
|---|---|---|---|---|---|---|
Accuracy | Precision | Recall | Specificity | F1-score | MCC | |
CNN | 84.3 | 83.2 | 82.3 | 81.4 | 82.4 | 0.82 |
VGG-19 | 74.3 | 73.2 | 72.1 | 70.4 | 72.5 | 0.72 |
DenseNETS | 93.2 | 90.5 | 90.3 | 89.5 | 90.4 | 0.88 |
RESNETS-34 | 91.4 | 91 | 90.4 | 90.2 | 90.7 | 0.90 |
Ensemble methods | 95.4 | 94.3 | 94.2 | 94 | 94.3 | 0.94 |
Resnets + CBAM | 93.2 | 93.4 | 92.4 | 93.0 | 92.9 | 0.92 |
Proposed model | 99.5 | 98.9 | 98.7 | 98.4 | 98.6 | 0.98 |