Hi Anderson,
Besides PALM for post hoc analysis, is there anything we can do for preprocessing to harmonize the data from different centers, so that we can correct for site effects partially before statistical analyses?
For example, for tbss_3_postreg, is it still recommended to use -S option to derive mean_FA and mean_FA_skeleton from mean of all subjects in study? Or for multicenter data, it's better to use -T option, using FMRIB58_FA and its skeleton to adjust the center variance?
Thanks,
Jian
|