Research Article
Classification on Digital Pathological Images of Breast Cancer Based on Deep Features of Different Levels
Table 14
Extracted features from different feature extraction nodes on image-level F1 score (/%).
| Point location | Classifier | Magnifications | 40× | 100× | 200× | 400× |
| 1 | SVM | 86.33 ± 2.05 | 87.17 ± 2.28 | 87.76 ± 2.05 | 84.28 ± 1.92 | LR | 82.80 ± 2.41 | 85.21 ± 5.32 | 84.93 ± 7.62 | 83.67 ± 4.42 | GNB | 74.81 ± 3.25 | 80.91 ± 2.67 | 82.54 ± 4.03 | 73.57 ± 4.86 | DT | 85.04 ± 1.42 | 85.69 ± 2.43 | 88.20 ± 1.27 | 86.08 ± 3.13 | RF | 83.69 ± 3.59 | 86.19 ± 2.16 | 88.52 ± 0.89 | 82.71 ± 6.64 | MFNN | 84.07 ± 2.74 | 85.15 ± 5.0 | 85.91 ± 4.43 | 84.62 ± 4.00 | 2 | SVM | 91.12 ± 3.77 | 93.30 ± 4.71 | 92.54 ± 2.77 | 90.45 ± 3.41 | LR | 87.97 ± 2.10 | 88.00 ± 2.93 | 88.73 ± 1.85 | 86.21 ± 1.51 | GNB | 86.61 ± 3.68 | 86.86 ± 2.95 | 89.62 ± 2.42 | 86.05 ± 5.27 | DT | 85.84 ± 2.71 | 86.74 ± 1.60 | 89.45 ± 1.86 | 86.23 ± 2.08 | RF | 84.21 ± 2.87 | 87.45 ± 2.00 | 90.76 ± 1.35 | 88.23 ± 5.47 | MFNN | 86.58 ± 2.40 | 87.33 ± 3.15 | 90.19 ± 2.41 | 87.67 ± 2.74 | 3 | SVM | 89.21 ± 0.95 | 90.98 ± 1.50 | 92.79 ± 2.66 | 89.83 ± 3.80 | LR | 87.34 ± 1.27 | 87.55 ± 1.10 | 88.69 ± 2.54 | 86.56 ± 3.85 | GNB | 85.12 ± 2.58 | 87.68 ± 1.61 | 89.55 ± 2.45 | 86.01 ± 5.00 | DT | 85.08 ± 1.71 | 86.27 ± 2.12 | 89.62 ± 3.91 | 86.23 ± 3.40 | RF | 84.80 ± 3.06 | 87.09 ± 1.48 | 90.04 ± 1.64 | 86.67 ± 3.21 | MFNN | 87.71 ± 1.40 | 88.98 ± 0.68 | 90.80 ± 2.75 | 85.91 ± 2.40 |
|
|