Table 2 Multi-classification results of comparison experiments based on the raw dataset (Raw) and augmented dataset (Aug).
From: Breast Cancer Multi-classification from Histopathological Images with Structured Deep Learning Model
Accuracy at | Methods | Magnification factors | |||
|---|---|---|---|---|---|
40X | 100X | 200X | 400X | ||
Image level | LeNet + Raw | 40.1 ± 7.1 | 37.5 ± 6.7 | 40.1 ± 3.4 | 38.2 ± 5.9 |
LeNet + Aug | 46.4 ± 4.5 | 47.34 ± 4.9 | 46.5 ± 5.6 | 45.2 ± 9.1 | |
AlexNet + Raw | 70.1 ± 7.4 | 68.1 ± 7.6 | 67.6 ± 4.8 | 67.3 ± 3.4 | |
AlexNet + Aug | 86.4 ± 3.1 | 75.8 ± 5.4 | 72.6 ± 4.8 | 84.6 ± 3.6 | |
CSDCNN + Raw | 89.4 ± 5.4 | 90.8 ± 2.5 | 88.6 ± 4.7 | 87.6 ± 4.1 | |
CSDCNN + Aug | 92.8 ± 2.1 | 93.9 ± 1.9 | 93.7 ± 2.2 | 92.9 ± 1.8 | |
Patient level | LeNet + Raw | 38.1 ± 9.3 | 37.5 ± 3.4 | 38.5 ± 4.3 | 37.2 ± 3.6 |
LeNet + Aug | 48.2 ± 4.5 | 47.6 ± 7.5 | 45.5 ± 3.2 | 45.2 ± 8.2 | |
AlexNet + Raw | 70.4 ± 6.2 | 68.7 ± 5.3 | 66.4 ± 4.3 | 67.2 ± 5.6 | |
AlexNet + Aug | 74.6 ± 7.1 | 73.8 ± 4.5 | 76.4 ± 7.4 | 79.2 ± 7.6 | |
CSDCNN + Raw | 88.3 ± 3.4 | 89.8 ± 4.7 | 87.6 ± 6.4 | 87.0 ± 5.2 | |
CSDCNN + Aug | 94.1 ± 2.1 | 93.2 ± 1.4 | 94.7 ± 3.6 | 93.5 ± 2.7 | |
