Skip to main content
. 2023 Feb 14;36(3):1262–1278. doi: 10.1007/s10278-022-00763-z

Table 1.

Comparison of different rigid registration methods. Standard deviation is provided in the brackets. The best performances are shown in bold

Methods ChestX-ray8
MSE MI Dice CPU time/s GPU time/s
Before 0.048 (0.017) 0.719 (0.206) 0.833 (0.076) - -
Elastix-affine 0.029 (0.020) 1.020 (0.194) 0.921 (0.030) 0.767 (0.055) -
ANTs-affine 0.032 (0.018) 1.038 (0.188) 0.922 (0.029) 0.622 (0.114) -
DLIR 0.041 (0.034) 0.915 (0.172) 0.914 (0.054) 0.025 (0.003) 0.005 (0.002)
GlobalNet 0.035 (0.041) 1.005 (0.198) 0.920 (0.034) 0.022 (0.003) 0.004 (0.003)
CCAB 0.028 (0.015) 1.032 (0.204) 0.926 (0.028) 0.025 (0.002) 0.018 (0.002)
Methods Test12
MSE MI Dice CPU time/s GPU time/s
Before 0.048 (0.012) 0.916 (0.199) 0.876 (0.045) - -
Elastix-affine 0.036 (0.021) 1.187 (0.103) 0.923 (0.045) 0.805 (0.041) -
ANTs-affine 0.038 (0.043) 1.235 (0.113) 0.924 (0.042) 0.627 (0.102) -
DLIR 0.043 (0.024) 1.088 (0.130) 0.917 (0.029) 0.024 (0.002) 0.005 (0.003)
GlobalNet 0.041 (0.021) 1.162 (0.120) 0.921 (0.029) 0.020 (0.002) 0.004 (0.002)
CCAB 0.030 (0.020) 1.256 (0.119) 0.928 (0.032) 0.027 (0.003) 0.006 (0.002)