Dear Shahrzad,
> I have performed one VBM analysis using VBM8 toolbox and SPM version
> (8.4010) with a Dartel template made based on my group. (And modulation:
> non-linear only)
>
> Now, I performed the same analysis but this time without the use of VBM8
> toolbox (subjects are the same) on a newer SPM version (8.4667).
> Meaning performing , new Segment ---> Dartel template creation --->
> Normalize to MNI (with Modulation)
>
> I used the code in the mailing list to calculate the Total Gray mater, WM
> and TIV from the C[13]* images of this analysis.
>
> Question 1)
> The global numbers that I have calculated here, differ from the global
> measures which the VBM8 toolbox gives as an output (rawvolumes.txt). Why??
The segmentation algorithms are rather different between the two
approaches, so you will get different results. In addition, the new
segment option includes more tissue classes, and so this alone will
change the total numbers if you are only considering the first 3
tissue classes (gray matter, white matter, csf), and not soft tissue
or bone.
We compared the old vs. new segment in our recent NeuroImage paper -
there are some plots in the supplemental material. We did not use the
VBM8 toolbox but the overall principle is similar.
> Question2)
> Now to perform the same statistics (two-sample t-test), I have added the
> calculated TIV once as a covariate and once as a global measure with option:
> Normalize : proportional. In both cases I get completely different results
> of the group differences in compare to the VBM8 results. No significant
> Voxel or cluster survives the corrections and even the uncorrected results
> are not so similar. One may be able to say that they have the same trend but
> it is far from being significant. What could cause this differences and what
> is the correct way for me to proceed?
The modulation in VBM8 that you have done - nonlinear only - will be
different than the SPM modulation. So, you have different segmentation
algorithms (see your first question), and now different types of
modulation. Thus, it's not surprising you get different answers.
For the differences between global scaling and a covariate approach,
we talk about this in the context of total gray matter in the
NeuroImage paper below. More specifically regarding TIV, you may want
to see Barnes et al (2010).
I don't think there is a good answer on the "correct" analysis
approach. It is worth thinking about the different assumptions
underlying the various options, and deciding which seems to match your
research question the best.
References:
Barnes J, Ridgway GR, Bartlett J, Henley SMD, Lehmann M, Hobbs N,
Clarkson MJ, MacManus DG, Ourselin S, Fox NC (2010) Head size, age and
gender adjustment in MRI studies: a necessary nuisance? NeuroImage
53:1244-1255.
Peelle JE, Cusack R, Henson RNA (2012) Adjusting for global effects in
voxel-based morphometry: Gray matter decline in normal aging.
NeuroImage 60:1503-1516.
Hope this helps!
Best regards,
Jonathan
--
Jonathan Peelle, PhD
Department of Otolaryngology
Washington University in St. Louis
Office: (314) 362-9044
http://peellelab.org || http://jonathanpeelle.net
|