Table 4 Segmentation results by data augmentation amount.
From: Laplacian filter attention with style transfer GAN for brain tumor MRI imputation
Method | Dataset | DICE | IoU | Precision | Recall |
|---|---|---|---|---|---|
Baseline | No Augmentation | 0.5691 | 0.5101 | 0.6707 | 0.6825 |
StarGAN v2 | 25% Augmentation | 0.6977 | 0.5398 | 0.6441 | 0.7669 |
50% Augmentation | 0.7034 | 0.5457 | 0.6355 | 0.7928 | |
CollaGAN | 25% Augmentation | 0.6817 | 0.5211 | 0.6195 | 0.7616 |
50% Augmentation | 0.7063 | 0.5491 | 0.6767 | 0.7428 | |
ResViT | 25% Augmentation | 0.6920 | 0.5465 | 0.6451 | 0.7673 |
50% Augmentation | 0.7256 | 0.5704 | 0.6882 | 0.7935 | |
LASTGAN | 25% Augmentation | 0.7110 | 0.5549 | 0.6572 | 0.7839 |
50% Augmentation | 0.7529 | 0.6086 | 0.6985 | 0.8265 |
