Table 1 Comparison results of the HyborNet method with current popular methods on the LiTS dataset.

From: Hybrid gabor attention convolution and transformer interaction network with hierarchical monitoring mechanism for liver and tumor segmentation

Methods

Liver

Tumor

DIC

IOU

Recall

RAVD

ASSD

MSD

DIC

IOU

Recall

RAVD

ASSD

MSD

DeepLabv3+

90.62

83.02

86.34

0.012

2.7

10.48

45.64

36.31

39.26

0.41

1.76

36.66

U-Net

94.64

90.02

91.12

0.013

2.25

12.97

53.84

40.35

47.35

1.57

7.66

30.92

Attention U-Net

94.56

89.89

91.43

0.012

1.99

9.04

53.74

40.29

47.42

0.68

9.08

39.99

ResU-Net

94.78

90.22

91.76

0.01

2.05

9.25

47.22

35.32

40.61

1.03

7.06

23.85

U-Net++

94.67

90.08

91.84

0.007

1.95

10.54

52.73

39.01

47.94

0.85

7.69

31.59

Double UNet

94.66

89.82

91.95

0.014

3.37

15.93

48.65

34.90

41.74

1.61

11.51

49.05

nnformer

94.61

90.14

92.32

0.009

3.23

18.23

49.42

35.25

45.65

1.59

11.94

37.65

Swin-UNet

94.18

90.37

92.48

-0.01

3.54

14.59

49.69

35.57

45.73

1.55

12.45

35.38

TransUNet

94.67

90.52

92.74

0.004

2.39

12.57

48.46

37.02

45.45

0.61

13.42

33.11

HiFormer

94.89

90.87

92.85

0.008

2.1

9.28

53.51

40.64

48.93

1.26

9.01

30.89

HyborNet

95.82

91.34

93.14

0.005

1.82

8.18

55.59

42.16

49.24

0.55

5.65

24.50

Back to article page