Dear Christian and SPM users, Thank you very much for your kind help! I am doing a VBM longitudinal analysis with three times, and data preprocessing follows the standard longitudinal preprocessing stream of CAT12. When I use ROI tools (estimate mean values inside ROI) of CAT12 to extract cerebellum volumes by lpba40 or suit atlas, I found some strange thing. In the middle time, the cerebellum gray matter volume had no change, while the cerebellum white matter volume decreased obviously compared to baseline. However, in the last time, the cerebellum gray matter volume decreased obviously, while the cerebellum white matter volume had no change compared to baseline. Since cerebellum atroph of participants of this study will progress gradually, this result seemed very strange. Moreover, when I check cerebellum volume data of each individual,I found that whose cerebellum gray matter volume decreased obviously, his cerebllum white matter volume increaseed obviously. When I use total cerebellum volume (gray matter+white matter) for analysis, it seems more reasonable, both cerebellum total volume of middle time and last time decreased, and the last time decreased more. So I doubt that maybe some cerebellum gray matter misclassified into white matter of some participants in the last time, and some white matter misclassified into gray matter in the middle time. So maybe total volume is more suitable? But many studies only analysis gray matter by VBM, so I wonder if this has happened to other people, and how to explain and solve it.
The ROI analysis by SPM.mat had similary result, areas which cerebellum gray matter volume decreased, the white matter volume increased. The whole brain analysis seemed to be less affected, the white matter volume in middle or last time both decreased, but the gray matter volume in middle time had little change, and changed obviously in the last time. I also want to ask if I use ROI volume for analysis, should I use TIV of each time point to normarlize it, or just use TIV in the baseline to normarlize it, or just use TIV in the baseline as a covariate?
This question really bothers me for a long time, and I looking forward to your kind help. I'm very sorry to trouble you.Thank you very much!
|