Table 4 Comparison of experimental outcomes among various networks on the ISIC2017 dataset.
From: Medical image segmentation model based on local enhancement driven global optimization
Methods | DSC\(\uparrow\) | HD\(\downarrow\) | Jaccard\(\uparrow\) | Precision\(\uparrow\) | Recall\(\uparrow\) |
|---|---|---|---|---|---|
U-Net4 | 83.06 | 10.35 | 74.15 | 94.31 | 78.79 |
R50 U-Net7 | 82.81 | 11.09 | 74.81 | 92.76 | 82.82 |
UNet++9 | 82.84 | 11.21 | 73.74 | 91.88 | 80.33 |
UNet3+10 | 89.85 | 1.33 | 88.76 | 86.07 | 94.75 |
R50 Att-UNet7 | 83.12 | 10.41 | 74.48 | 93.41 | 79.57 |
CBAM36 | 83.08 | 10.02 | 74.13 | 94.96 | 78.43 |
SENet37 | 82.89 | 10.23 | 73.92 | 95.11 | 77.89 |
SKNet38 | 82.97 | 10.06 | 73.95 | 95.23 | 78.01 |
Att-UNet8 | 83.12 | 10.41 | 74.48 | 93.41 | 79.57 |
TransUNet19 | 83.25 | 10.01 | 74.43 | 94.41 | 79.18 |
MT-UNet21 | 72.57 | 18.04 | 61.58 | 85.96 | 71.84 |
TransClaw23 | 83.49 | 9.32 | 74.72 | 95.00 | 78.77 |
SwinUNet32 | 80.38 | 12.26 | 70.86 | 91.17 | 77.75 |
TransDeeplab41 | 84.00 | 8.42 | 75.14 | 93.38 | 80.57 |
LEGO-Net(Ours) | 87.88 | 6.44 | 79.99 | 89.50 | 89.69 |
