Table 3 Quantitative comparison of state-of-the-art methods on the ISIC benchmark
From: CFG-MambaNet: Contextual and Frequency-Guided Mamba Network for medical image segmentation
Model | Dice (%) ↑ | IoU (%) ↑ | ASD ↓ | Recall (%) ↑ |
|---|---|---|---|---|
U-Net32 | 91.56 | 85.95 | 5.19 | 94.83 |
nnUNet33 | 93.11 | 88.03 | 4.17 | 96.30 |
AttUNet12 | 92.20 | 86.79 | 4.38 | 94.69 |
MISSFormer37 | 92.52 | 87.10 | 4.19 | 96.06 |
FSCA-Net40 | 93.21 | 88.19 | 3.60 | 94.84 |
Rolling-unet34 | 84.13 | 77.47 | 12.59 | 94.25 |
H2Former36 | 92.80 | 88.47 | 7.50 | 95.56 |
UCTransNet35 | 89.42 | 82.79 | 7.06 | 94.84 |
EMCAD41 | 93.74 | 88.95 | 3.30 | 96.41 |
Hetero-UNet38 | 91.57 | 85.77 | 5.47 | 95.18 |
GH-UNet42 | 93.74 | 89.10 | 5.58 | 96.69 |
Swin-UMamba39 | 93.99 | 89.32 | 3.55 | 96.82 |
Ours | 94.46 | 89.83 | 3.42 | 97.71 |
