Table 5 Segmentation accuracy performance of the proposed tooth instance segmentation method on CBCT images from different sources

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

Manufacturer

Manufacturer’s model name

Dice/%

mIoU/%

HD/mm

ASD/mm

Carestream Health

CS 9300, CS 9301

95.9 ± 1.2

89.1 ± 1.0

1.45 ± 0.23

0.17 ± 0.25

Imaging Sciences International

9–17

95.3 ± 1.2

88.6 ± 0.2

1.42 ± 0.21

0.18 ± 0.11

J.Morita.Mfg.Corp.

 

96.2 ± 0.8

90.3 ± 0.9

1.15 ± 0.31

0.09 ± 0.08

LargeV

HighRes3D, SMART3D

96.9 ± 0.3

91.2 ± 0.8

1.22 ± 0.25

0.09 ± 0.10

NewTom

NTVGiMK4, NTVGiEVO, NT5G

93.8 ± 1.5

86.7 ± 0.6

1.10 ± 0.25

0.16 ± 0.13

NNT

NTVGiEVO

96.1 ± 1.2

89.8 ± 0.9

1.22 ± 0.42

0.15 ± 0.32

PaloDEx Group Oy

ORTHOPANTOMOGRAPH OP 3D

94.8 ± 0.9

87.9 ± 1.1

1.81 ± 0.21

0.18 ± 0.23

RAY Co., Ltd.

RAYSCAN N Alpha Plus

95.9 ± 1.1

89.4 ± 0.7

1.84 ± 0.25

0.15 ± 0.14

Sirona

ORTHOPHOS SL

94.1 ± 2.1

86.8 ± 1.1

2.72 ± 0.62

0.42 ± 0.52

Vatech Company Limited

PHT-35LHS

93.7 ± 3.8

85.5 ± 2.0

2.12 ± 0.40

0.32 ± 0.24

YOFO

Pirox-R

96.1 ± 1.2

90.1 ± 0.8

1.21 ± 0.35

0.16 ± 0.09

  1. Bold text represents the highest value in its column