Table 2 Experimental results of quantitative comparison with existing advanced methods in terms of segmentation and detection accuracy

From: Fully automatic AI segmentation of oral surgery-related tissues based on cone beam computed tomography images

Methods

Dice/%

mIoU/%

HD/mm

ASD/mm

Hi-MoToothSeg

93.1 ± 0.8

82.5 ± 1.8

1.63 ± 0.75

0.28 ± 0.14

nnUNet

85.3 ± 2.5

75.1 ± 2.0

5.04 ± 2.48

0.51 ± 0.31

ToothNet

91.7 ± 1.3

76.2 ± 0.7

2.85 ± 1.11

0.49 ± 0.08

RELU-Net

92.9 ± 1.0

85.7 ± 0.9

1.52 ± 0.42

0.24 ± 0.11

DenseASPP-UNet

92.5 ± 1.4

79.4 ± 1.2

2.34 ± 0.76

0.31 ± 0.21

Ours

94.3 ± 1.0

86.3 ± 1.1

1.43 ± 0.52

0.18 ± 0.04

  1. Bold text represents the highest value in its column
Back to article page