Table 3.
The consistencies of regional brain SUVs among the three pairs of segmentation strategies at the whole-brain and cerebral cortex level
The consistencies among different methods | With or without MRI registration | With atlas- or individual-based algorithms | With or without AI | |||||
---|---|---|---|---|---|---|---|---|
With atlas-based algorithms | With individual-based algorithms | All | With MRI registration | Without MRI registration | All | |||
Whole-brain level | ICC | 0.984 | 0.951 | 0.985 | 0.963 | 0.936 | 0.970 | 0.998 |
LB | 0.962 | 0.884 | 0.964 | 0.913 | 0.851 | 0.928 | 0.995 | |
UB | 0.994 | 0.980 | 0.994 | 0.985 | 0.973 | 0.987 | 0.999 | |
Cerebral cortex level | ICC | 0.974 | 0.924 | 0.975 | 0.741 | 0.879 | 0.819 | 0.997 |
LB | 0.954 | 0.869 | 0.956 | 0.582 | 0.795 | 0.699 | 0.995 | |
UB | 0.985 | 0.957 | 0.986 | 0.846 | 0.930 | 0.894 | 0.998 |
ICC intraclass correlation coefficients, UB and LB the upper bound and lower bound of 95% confidential interval