Table 8 The effect of different positions of ASPP and SE-Attention modules on the Synapse dataset.

From: MedFuseNet: fusing local and global deep feature representations with hybrid attention mechanisms for medical image segmentation

Position

Average DSC

HD

Aorta

Gallbladder

Kidney(L)

Kidney(R)

Liver

Pancreas

Spleen

Stomach

A[\(A_2\)-\(A_3\)] + S[\(B_1\)-\(B_2\)]

78.40

18.44

85.71

62.58

85.10

77.76

93.98

58.16

75.33

78.74

A[\(A_2\)-\(A_3\)] + S[\(B_0\)-\(B_1\)]

77.94

32.68

84.75

63.21

85.00

79.01

93.86

56.46

89.39

70.49

A[\(A_2\)-\(A_3\)] + S[\(B_2\)-\(B_3\)]

76.03

33.24

84.84

62.70

77.99

73.62

92.45

59.84

83.12

73.72

A[\(A_3\)-\(A_4\)] + S[\(B_1\)-\(B_2\)]

74.75

16.75

85.02

55.02

84.28

81.15

94.36

43.84

82.59

71.72

A[\(A_1\)-\(A_2\)] + S[\(B_1\)-\(B_2\)]

74.61

26.04

84.23

56.12

77.31

72.04

93.08

55.59

85.60

72.89

  1. 1. ’A’ represents the ASPP module, ’S’ denotes the SE-Attention module, \(A_1\), \(A_2\), \(A_3\), and \(A_4\) denote the position in the encoder, whereas \(B_1\), \(B_2\), and \(B_3\) is the position in the decoder
Back to article page