Table 2 Evaluation indicators ± standard deviation of all competing methods in the Skin Lesion Segmentation.

From: Dual-branch dynamic hierarchical U-Net with multi-layer space fusion attention for medical image segmentation

Methods

DICE (%)

IOU (%)

RAVD (%)

ASSD

MSSD

FCN

84.92 ± 0.39

78.61 ± 0.45

13.22 ± 0.45

14.47 ± 4.6

42.77 ± 3.5

U-Net

85.53 ± 0.47

79.83 ± 0.50

14.14 ± 0.36

13.44 ± 3.5

38.77 ± 4.2

ResU-Net

87.87 ± 0.44

80.27 ± 0.69

12.91 ± 0.43

13.27 ± 4.1

37.90 ± 3.6

Attention U-Net

87.68 ± 0.35

79.94 ± 0.43

13.32 ± 0.31

13.52 ± 4.7

38.41 ± 2.7

U-Net++

87.74 ± 0.36

80.06 ± 0.25

13.78 ± 0.28

13.23 ± 3.1

38.37 ± 2.2

ResU-Net++

87.39 ± 0.25

79.51 ± 0.21

14.11 ± 0.31

13.75 ± 2.7

39.08 ± 1.8

TransformU-Net

88.91 ± 0.19

81.75 ± 0.18

− 17.96 ± 0.09

11.21 ± 1.6

50.57 ± 1.7

SwimU-Net

88.26 ± 0.16

80.85 ± 0.21

11.22 ± 0.12

11.80 ± 1.8

31.55 ± 2.1

HiFormer

88.93 ± 0.29

81.89 ± 0.25

15.21 ± 0.13

11.41 ± 2.1

31.74 ± 2.5

D2HU-Net

90.18 ± 0.17

82.01 ± 0.15

5.74 ± 0.21

18.83 ± 1.2

28.96 ± 2.5

Back to article page