Table 3. Test accuracy of scan-based classification obtained by aggregating patch-based classification and determining the most frequent prediction.
| Method | CA | CG | CL | CN | CP | CT | MF | SB | SC | Total |
|---|---|---|---|---|---|---|---|---|---|---|
| AlexNet | 95.5 ± 5.0 | 95.0 ± 5.0 | 70.0 ± 10.0 | 57.0 ± 6.9 | 95.0 ± 5.0 | 45.0 ± 10.0 | 70.0 ± 10.0 | 75.0 ± 5.0 | 85.0 ± 5.0 | 77.3 ± 4.2 |
| DenseNet169 | 80.0 ± 10.0 | 85.0 ± 5.0 | 45.0 ± 5.0 | 57.0 ± 21.2 | 70.0 ± 10.0 | 100.0 ± 0.0 | 85.0 ± 15.0 | 95.0 ± 5.0 | 75.0 ± 10.0 | 77.6 ± 6.6 |
| InceptionV3 | 50.0 ± 10.0 | 50.0 ± 10.0 | 80.0 ± 0.0 | 78.6 ± 14.3 | 55.0 ± 10.0 | 60.0 ± 10.0 | 50.0 ± 10.0 | 85.0 ± 5.0 | 85.0 ± 5.0 | 65.9 ± 4.9 |
| ResNet18 | 100.0 ± 0.0 | 75.0 ± 5.0 | 100.0 ± 0.0 | 78.6 ± 6.9 | 50.0 ± 10.0 | 70.0 ± 0.0 | 70.0 ± 10.0 | 95.0 ± 5.0 | 80.0 ± 10.0 | 78.3 ± 5.4 |
| ResNet50 | 100.0 ± 0.0 | 85.0 ± 15.0 | 100.0 ± 0.0 | 57.0 ± 6.9 | 50.0 ± 10.0 | 45.0 ± 15.0 | 80.0 ± 10.0 | 95.0 ± 5.0 | 75.0 ± 5.0 | 78.1 ± 8.3 |
| AlexNet BoW RF | 100.0 ± 0.0 | 75.0 ± 25.0 | 100.0 ± 0.0 | 75.0 ± 15.0 | 100.0 ± 0.0 | 90.0 ± 10.0 | 90.0 ± 10.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 92.2 ± 4.4 |
| InceptionV3 BoW RF | 80.0 ± 20.0 | 75.0 ± 25.0 | 100.0 ± 0.0 | 45.0 ± 15.0 | 40.0 ± 0.0 | 60.0 ± 30.0 | 0.0 ± 0.0 | 70.0 ± 10.0 | 45.0 ± 15.0 | 57.2 ± 2.8 |
| ResNet18 BoW RF | 100.0 ± 0.0 | 70.0 ± 10.0 | 100.0 ± 0.0 | 45.0 ± 15.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 80.0 ± 10.0 | 90.0 ± 10.0 | 100.0 ± 0.0 | 87.2 ± 1.7 |
| AlexNet BoW SVM | 100.0 ± 0.0 | 65.0 ± 25.0 | 100.0 ± 0.0 | 70.0 ± 10.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 85.0 ± 5.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 91.1 ± 2.2 |
| InceptionV3 BoW SVM | 70.0 ± 10.0 | 55.0 ± 15.0 | 100.0 ± 0.0 | 40.0 ± 20.0 | 45.0 ± 5.0 | 55.0 ± 15.0 | 0.0 ± 0.0 | 55.0 ± 15.0 | 40.0 ± 20.0 | 51.1 ± 2.3 |
| ResNet18 BoW SVM | 100.0 ± 0.0 | 60.0 ± 20.0 | 100.0 ± 0.0 | 65.0 ± 0.0 | 100.0 ± 0.0 | 90.0 ± 10.0 | 60.0 ± 0.0 | 95.0 ± 5.0 | 100.0 ± 0.0 | 85.6 ± 2.2 |
| AlexNet FV RF | 100.0 ± 0.0 | 65.0 ± 35.0 | 100.0 ± 0.0 | 55.0 ± 5.0 | 100.0 ± 0.0 | 90.0 ± 10.0 | 95.0 ± 5.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 89.4 ± 2.2 |
| InceptionV3 FV RF | 65.0 ± 5.0 | 95.0 ± 5.0 | 100.0 ± 0.0 | 50.0 ± 10.0 | 30.0 ± 10.0 | 75.0 ± 25.0 | 5.0 ± 5.0 | 45.0 ± 25.0 | 45.0 ± 35.0 | 56.7 ± 3.3 |
| ResNet18 FV RF | 95.0 ± 5.0 | 60.0 ± 0.0 | 100.0 ± 0.0 | 65.0 ± 5.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 95.0 ± 5.0 | 95.0 ± 5.0 | 95.0 ± 5.0 | 89.4 ± 1.7 |
| AlexNet FV SVM | 100.0 ± 0.0 | 75.0 ± 25.0 | 100.0 ± 0.0 | 75.0 ± 15.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 95.0 ± 5.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 93.9 ± 3.9 |
| InceptionV3 FV SVM | 75.0 ± 25.0 | 60.0 ± 0.0 | 100.0 ± 0.0 | 55.0 ± 5.0 | 45.0 ± 15.0 | 85.0 ± 5.0 | 5.0 ± 5.0 | 25.0 ± 15.0 | 45.0 ± 25.0 | 55.0 ± 5.6 |
| ResNet18 FV SVM | 100.0 ± 0.0 | 60.0 ± 0.0 | 100.0 ± 0.0 | 45.0 ± 15.0 | 95.0 ± 5.0 | 100.0 ± 0.0 | 95.0 ± 5.0 | 100.0 ± 0.0 | 100.0 ± 0.0 | 88.3 ± 2.7 |