Table 2 Segmentation results on ACDC dataset. All methods are evaluated with DSC metric (%).
From: Multi-scheme cross-level attention embedded U-shape transformer for MRI semantic segmentation
Methods | MYO | RV | LV | Mean DSC (%) |
|---|---|---|---|---|
R50 U-Net9 | 80.63 | 87.10 | 94.92 | 87.55 |
R50 Att-UNet9 | 79.20 | 87.58 | 93.47 | 86.75 |
R50 ViT9 | 81.88 | 86.07 | 94.75 | 87.57 |
TransUnet9 | 84.53 | 88.86 | 95.73 | 89.71 |
SwinUNet8 | 85.62 | 88.55 | 95.83 | 90.00 |
TransUNet+38 | 87.98 | 89.12 | 94.12 | 90.42 |
MT-UNet39 | 89.04 | 86.64 | 95.62 | 90.43 |
PVT-CASCADE40 | 89.97 | 88.90 | 95.50 | 91.46 |
TransCASCADE40 | 90.25 | 89.14 | 95.50 | 91.63 |
MAXFormer35 | 90.89 | 89.53 | 96.02 | 92.15 |
MSCL-SwinUNet (ours) | 91.33 | 89.75 | 95.91 | 92.33 |
