Dear Liang,
the reason to skip the modulation step for longitudinal data was that the effect of modulation on longitudinal data is rather subtle. Because the deformations due to non-linear normalization are the same for all time points the modulation only depends on the affine component that is driven by different brain sizes. You could try to define a covariate for total brain size (TIV) in your design to covariate out this potential effect. However, I don't expect too much change of your results...
Best,
Christian
On Thu, 5 Dec 2013 16:24:22 +0800, Liang Wang <[log in to unmask]> wrote:
>Hi Christian and SPM folks,
>
>I am working on a longitudinal data using VBM8. Following the VBM-manual
>and changing the input data format to adapt to a longitudinal analysis, I
>sucessfully got the warped normalized gray matter images (wp1mr*). By
>searching this mail list, I saw that a long time ago Christian mentioned
>that it was not necessary to output the modulated GM volume (rather than GM
>intensity) because the spatial normalization (= inverse modulation) is the
>same across multiple time points. It sounds reasonable. However, I still
>feel this measure (GM intensity, usually less than 1) was hard to be
>understood and the results were also hardly interpreted. I am wondering how
>to make VBM8 to output the modulated images for the longitudinal analysis.
>I tried to change the parameters stored in cg_vbm8_defaults.m, but it did
>not work. Thanks.
>
>Best,
>Liang
>
>--
>Liang Wang, PhD
>Institute of Psychology
>Chinese Academy of Sciences
>Beijing 100101, China
>
|