Early Stopping and 5-Fold Cross-Validation for Optimized Model Training

We used an optimizer for 300 epochs. When the AUC did not increase for 30 epochs, we stopped the training early and used the model with the best AUC. The initial learning rate was 0.001, and the weight decay was 0.01. We randomly split the dataset into a training set and a validation set in a 4:1 ratio and used 5-fold cross-validation to evaluate the model. For all backbone networks, we used pre-trained weights on ImageNet.