Table 9 Performance comparison between our scheme (with assembling) and the state-of-the-art work [11] on BACH
From: Breast cancer histopathology image classification through assembling multiple compact CNNs
Methods | Strategy | PL | IL | Kappa | PPV |
|---|---|---|---|---|---|
Work [11] | 32 ×32 (random sampling) | 80.5 | 76.8 | 0.608 | 78.7 |
|  | 64 ×64 (random sampling) | 79.9 | 74.8 | 0.595 | 76.5 |
|  | 32 ×32 (sliding window) | 80.4 | 75.5 | 0.607 | 76.7 |
|  | 64 ×64 (sliding window) | 82.1 | 77.1 | 0.641 | 76.2 |
| Â | Max Fusion | 85.1 | 79.3 | 0.700 | 78.4 |
This work | Hybrid model with assembling (Sum) | 87.5 | 84.4 | 0.749 | 85.7 |
| Â | Hybrid model with assembling (Max) | 87.4 | 84.2 | 0.748 | 84.6 |