Table 4.
Breast cancer classification results and comparison for binary classification (benign vs. malignant tumor) with data augmentation and without data augmentation techniques on the BreakHis dataset
| Method | Year | Classification rate at magnification factor | ||||
|---|---|---|---|---|---|---|
| ×40 | ×100 | ×200 | ×400 | |||
| Image level | CNN + fusion(sum, product, max) [15] | 2016 | 85.6 ± 4.8 | 83.5 ± 3.9 | 83.6 ± 1.9 | 80.8 ± 3.0 |
| AlexNet + Aug. [18] | 2017 | 85.6 ± 4.8 | 83.5 ± 3.9 | 83.1 ± 1.9 | 80.8 ± 3.0 | |
| ASSVM [19] | 94.97 | 93.62 | 94.54 | 94.42 | ||
| CSDCNN + Aug [18] | 2017 | 95.80 ± 3.1 | 96.9 ± 1.9 | 96.7 ± 2.0 | 94.90 ± 2.8 | |
| IRRCNN +without Aug. | 2018 | 97.16 ± 1.37 | 96.84 ± 1.34 | 96.61 ± 1.31 | 95.78 ± 1.44 | |
| IRRCNN + Aug. | 2018 | 97.95 ± 1.07 | 97.57 ± 1.05 | 97.32 ± 1.22 | 97.36 ± 1.02 | |
| Patient level | CNN + fusion (sum, product, max) [15] | 2016 | 90.0 ± 6.7 | 88.4 ± 4.8 | 84.6 ± 4.2 | 86.10 ± 6.2 |
| Bayramoglu et al. [14] | 2016 | 83.08 ± 2.08 | 83.17 ± 3.51 | 84.63 ± 2.72 | 82.10 ± 4.42 | |
| Multi-classifier by Gupta et al. [13] | 2017 | 87.2 ± 3.74 | 88.22 ± 3.23 | 88.89 ± 2.51 | 85.82 ± 3.81 | |
| CSDCNN + Aug. [18] | 2017 | 92.8 ± 2.1 | 93.9 ± 1.9 | 93.7 ± 2.2 | 92.90 ± 2.7 | |
| IRRCNN +without Aug. | 2018 | 96.69 ± 1.18 | 96.37 ± 1.29 | 96.27 ± 1 .57 | 96.15 ± 1.61 | |
| IRRCNN + Aug. | 2018 | 97.60 ± 1.17 | 97.65 ± 1.20 | 97.56 ± 1.07 | 97.62 ± 1.13 | |