Table 4.
Comparison between our developed GAN and other models in previous studies for the classification of leukemias.
| Classification Model | Tested Data Set | Classification Task | Accuracy (%) |
|---|---|---|---|
| CNN [34] | ALL-IDB and ASH image bank | Binary (ALL vs. normal) | 88.25 |
| Multi-class (acute and chronic leukemia sub-types) | 81.74 | ||
| SVM [55] | ASH image bank | Binary (AML vs. normal) | 98.00 |
| VGG-16 [54] | ALL-IDB | Binary (ALL vs. normal) | 96.84 |
| DenseNet-121 [4] | Private Dataset from Guangdong Second Provincial General | Multi-Class (ALL, AML, CML, and Normal) | 95.30 |
| Hospital, and Zhujiang Hospital of Southern Medical University | |||
| DenseNet-121 with SVM ResNet-50 with SVM [8] | Mixed data set including ALL-IDB | Binary (ALL vs. Normal) | 98.00 |
| images | Multi-class (ALL, AML, and Normal) | 96.67 | |
| Developed GAN Classifier | ALL-IDB and ASH image bank | Binary (ALL vs. Normal) | 98.65 |
| Multi-class (ALL, AML, and Normal) | 95.58 |