Table 12 Performance analysis of existing and proposed models on MIAS dataset using different optimizers after data-preprocessing (before data preprocessing).
CNN classifier | Optimizer | Accuracy (%) | Sensitivity (%) | Specificity (%) | Precision (%) | AUC (%) | F1-score (%) | Cohen’s Kappa (κ) |
|---|---|---|---|---|---|---|---|---|
VGG-16 | Adam | 70.50 | 61.00 | 82.50 | 68.00 | 78 | 64.35 | 0.51 |
RMSProp | 69.00 | 60.50 | 81.50 | 67.00 | 76 | 63.50 | 0.50 | |
SGD | 68.00 | 59.00 | 80.00 | 65.50 | 75 | 62.00 | 0.48 | |
VGG-19 | Adam | 75.50 | 66.50 | 85.00 | 71.50 | 80 | 71.00 | 0.54 |
RMSProp | 73.00 | 64.00 | 83.00 | 69.00 | 78 | 68.50 | 0.52 | |
SGD | 71.50 | 63.00 | 82.50 | 68.00 | 76 | 66.50 | 0.50 | |
DenseNet169 | Adam | 74.00 | 64.50 | 84.00 | 70.50 | 79 | 68.80 | 0.53 |
RMSProp | 72.50 | 63.00 | 83.50 | 69.00 | 78 | 67.50 | 0.51 | |
SGD | 71.00 | 62.00 | 81.00 | 68.00 | 77 | 65.50 | 0.49 | |
ResNet-50 | Adam | 76.50 | 67.50 | 86.50 | 72.50 | 81 | 70.70 | 0.55 |
RMSProp | 74.50 | 65.50 | 85.00 | 71.00 | 79 | 69.50 | 0.53 | |
SGD | 73.00 | 64.00 | 83.50 | 70.00 | 78 | 68.00 | 0.51 | |
DenseNet201 | Adam | 78.00 | 69.00 | 88.00 | 73.50 | 82 | 71.50 | 0.56 |
RMSProp | 76.50 | 68.00 | 87.50 | 72.00 | 81 | 70.60 | 0.54 | |
SGD | 75.00 | 67.50 | 86.00 | 71.50 | 79 | 69.00 | 0.52 | |
Proposed hybrid model | Adam | 92.00 | 85.00 | 93.50 | 90.00 | 95 | 88.50 | 0.75 |
RMSProp | 90.50 | 83.50 | 92.00 | 89.00 | 94 | 87.50 | 0.73 | |
SGD | 88.00 | 80.00 | 91.00 | 86.00 | 92 | 84.50 | 0.70 |
