Currently, the medical image classification is challenged by performance degradation due to imbalanced data. Balancing the data through sample augmentation proves to be an effective solution. However, ...