Table 3 Comparative analysis of brain tumor classification (BTC) methods: evolution and performance metrics.
References | Method | Number of classes | Precision | Recall val | F1_val | Accuracy val |
|---|---|---|---|---|---|---|
Multi scale CNN | 4 | 90.95% | 91.05% | 91% | 91.20% | |
Resnet 50 | 2 | – | – | – | 95% | |
Xception | 4 | 95.84% | 95.60% | 95.72% | 95.87% | |
SVM | 3 | 90.70% | 90.10% | 90.20% | 90.27% | |
VGG 19 | 3 | 94% | 94% | 94% | 94% | |
2D CNN | 4 | – | – | – | 93.44% | |
CNN NADE | 3 | 95% | 95% | 95% | 95% | |
SVM | 2 | – | – | – | 97% | |
CNN SVM | 3 | 98% | 97.90% | 97.90% | 95.82% | |
CNN | 4 | 95.32% | 95% | 95.36% | 95.44% | |
SVM with HOG, LBP and PCA | 4 | 96.02% | 96.03% | 96% | 96.03% | |
|  | Proposed Model (VGG16 + Custom Attention + Grad-CAM) | 4 | 98.97% | 98.94% | 98.95% | 99.00% |
