Table 1 Segmentation results of compare models on the Synapse dataset.

From: Medical image segmentation by combining feature enhancement Swin Transformer and UperNet

Model

PA (%)

Dice (%)

HD

Parameters(M)

Inference Time(s)

ResNet50+DeepLabV3+38

87.24

85.75

29.34

44.5

0.157

ViT+UPernet39

90.65

87.57

23.25

40.6

0.246

TransUNet40

91.67

89.13

20.58

78

0.309

SUnet41

86.23

84.29

13.22

20.9

0.122

VM-UNet42

85.58

81.08

19.21

44.27

0.136

FE-SwinUper

92.65

91.58

12.96

55

0.198

Back to article page