Skip to main content
. Author manuscript; available in PMC: 2021 Feb 1.
Published in final edited form as: Med Image Anal. 2019 Nov 23;60:101621. doi: 10.1016/j.media.2019.101621

Table 1:

Results achieved by 8 different methods in bone segmentation (i.e., midface, and mandible) and landmark digitization (i.e., average error for 15 landmarks).

Bone Segmentation Landmark

Methods Midface Mandible Digitization

DSC (%) SEN (%) PPV (%) DSC (%) SEN (%) PPV (%) Average Error (mm)

MA 81.14 ± 2.54 80.17 ± 3.27 82.48 ± 2.85 83.82 ± 2.21 84.31 ± 2.21 83.29 ± 2.30 3.05 ± 1.54
RF 86.86 ± 1.63 87.36 ± 2.98 85.92 ± 2.28 88.21 ± 1.52 88.54 ± 2.77 88.01 ± 1.95 2.67 ± 1.58
LSR 92.27 ± 1.31 91.96 ± 2.86 92.64 ± 1.83 89.19 ± 1.75 89.55 ± 2.34 89.03 ± 1.62 -
SPRF - - - - - - 1.52 ± 1.25

JSD1 91.83 ± 1.06 90.05 ± 2.35 93.72 ± 1.24 91.66 ± 1.07 91.35 ± 2.13 91.99 ± 1.01 1.78 ± 1.31
JSD2 92.20 ± 1.02 92.73 ± 2.50 91.78 ± 2.14 92.17 ± 0.99 93.30 ± 2.29 91.13 ± 1.41 1.33 ± 0.92
JSD3 91.89 ± 1.15 91.50 ± 2.63 92.02 ± 1.99 92.03 ± 1.08 93.14 ± 2.17 91.15 ± 1.52 1.49 ± 1.28

JSD 93.19 ± 0.89 92.82 ± 1.91 93.61 ± 1.40 93.27 ± 0.97 93.63 ± 1.37 92.93 ± 1.09 1.10 ± 0.71