Dear Carlos,
In my fMRI higher level analysis using simple regression (performed
in SPM 5), I found that activations in brain regions A and B to a
task were negatively correlated with age. Since gray matter density
(GMD) in some brain regions is also negatively correlated with the
age, some may argue that reduced activations in both regions may be
due to tissue loss rather than BOLD reduction to the task. To prove
aging effects on BOLD responses, I have to make correction for GMD
loss. As such, I performed another VBM analysis (in VBM 5.1) to get
the picture of GMD changes.
My question is, what then should I do to correct BOLD responses for
brain tissue loss?
There is another angle to this question which has to do with how accurate the spatial normalization is. If spatial normalization were perfect, the reduced gray matter volume in older adults wouldn't (directly) matter, because it would effectively be "expanded" during normalization (given the typical preprocessing step of not adjusting fMRI images during normalization). I think this is one reason that this issue is sometimes not an area of focus. However, it still may be worth investigating, and both of your suggestions seem reasonable to me.
I have 2 ideas but each has something I don’t know how to perform:
(1) In the fMRI higher-level simple regression analysis using age as
covariate of interest, use each subject’s VBM data as covariate of no
interest, factoring out it’s effect. But how can I do this? How to
put each subject’s VBM data (file sm0w….?) in the design matrix?
The biological parametric mapping (BPM) toolbox will allow the specification of voxelwise covariates as you suggest:
http://fmri.wfubmc.edu/software/Bpm(2) Produce 2 ROIs from fMRI group analysis representing brain
regions A and B. Use them to extract individual VBM data in these 2
regions. In SPSS, perform a partial correlation analysis (variables:
BOLD signals in A/B, and age) controlling for regional VBM data. But
how can I extract individual VBM data in these 2 regions?
Extracting the gray matter volume is straightforward, just like extracting values from fMRI data. You could try using MarsBar, for example:
http://marsbar.sourceforge.net/
I don't think there is a consensus on what the "best" answer to this question is - including gray matter volume, or not, is really asking two different questions. In my experience it is not commonly done, but that doesn't mean that it is not informative.
Hope this helps!
Best regards,
Jonathan
--
Dr. Jonathan Peelle
Center for Cognitive Neuroscience and
Department of Neurology
University of Pennsylvania
3 West Gates
3400 Spruce Street
Philadelphia, PA 19104
USA
http://jonathanpeelle.net/