Table 1.
Performance of three CNN models for three different patch sizes
Patch sizes | F-score (%) | Accuracy (%) | Sensitivity (%) | AUC (%) | |
---|---|---|---|---|---|
The first CNN model (M1) | 64 × 64 | 75.09 | 81.22 | 75.71 | 79.99 |
32 × 32 | 78.54 | 83.8 | 81.26 | 84.66 | |
16 × 16 | 81.09 | 85.65 | 80.01 | 82.74 | |
The second CNN model (M2) | 64 × 64 | 81.84 | 86.41 | 83.12 | 85.64 |
32 × 32 | 83.45 | 88.28 | 83.82 | 87.37 | |
16 × 16 | 84.3 | 87.03 | 82.35 | 85.65 | |
The third CNN model (M3) | 64 × 64 | 81.92 | 86.35 | 82.31 | 85.36 |
32 × 32 | 82.79 | 86.84 | 83.67 | 86.73 | |
16 × 16 | 83.53 | 87.504 | 82.49 | 86.17 |
The rows marked in italics illustrate the patch size achieving the highest sensitivity and AUC in the specific model