Skip to main content

Table 1 Quantitative comparisons of landmark detection performance with different detection networks using successful detection rate (SDR) and mean radial error (MRE)

From: Ceph-Net: automatic detection of cephalometric landmarks on scanned lateral cephalograms from children and adolescents using an attention-based stacked regression network

Models

SDR (%)

MRE

(mm)

1.0

2.0

3.0

4.0

5.0

U-Net

36.40

68.56

83.15

90.10

93.90

2.21 ± 6.62

SegNet

16.78

50.72

74.75

86.46

91.88

2.66 ± 4.50

Dense U-Net

36.85

69.55

83.18

90.60

94.51

1.94 ± 3.64

Attention U-Net

32.57

65.47

80.89

89.17

93.67

2.08 ± 3.17

Ceph-Net

41.35

73.14

85.22

91.18

94.65

1.75 ± 1.67