VBM and DTI provide different interpretations of the data. VBM tells
you about volume differences between groups or over time, while DTI
tells you about the integrity of the remaining tissue. That is to say,
if you lose, for example the fibers from the LGN to V1 you would have
less white matter (not tested), but the MD/FA values will not be
different than normal as DTI compares the existing tissue and not
tissue that has been entirely lost (Shimony et al. 2006; PMID:
16400157). Likewise, it has been proposed that DTI will predict
atrophy in AD and that regions that have already atrophied won't show
DTI differences as DTI compares the remaining tissue and differences
indicates indicate susceptible tissue (Canu et al. 2010; PMID:
20157252).
Based on these papers, doing both DTI and VBM on white matter are worthwhile.
On Friday, December 17, 2010, Jonathan Peelle <[log in to unmask]> wrote:
> Hi Ben
>
>> 1. Is normalizing to mni-Space reasonable in Schizophrenic patients (given that recent Meta-analyses indicate Morphologic differences between Schizophrenic & normal subjects)?
>
> I wouldn't expect this to be a problem. The template produced by
> DARTEL, which is used to estimate the transformation to MNI space,
> will include both people with schizophrenia and controls, and I
> suspect will be similar enough that you would get a reasonable
> normalization to MNI space. In general people seem to have success
> with all sorts of populations who may differ from the MNI-space
> template (dementia patients, older adults, children...), so it should
> be fine. If anyone has had difficulty with this step, though, it
> would be good to hear about it...
>
>
>> 2. Regarding the statical analysis: Is it valid to restrict the group-comparison of Gray matter differences to the Gray matter by restricting (masking) the Analysis to Grey matter?
>
> Presumably for this analysis, the images you are putting in are the
> gray matter segmentations. Thus, any results you get are already in
> some sense restricted to gray matter. But because the segmentations
> aren't always perfect, there are things that crop up outside, and so
> some sort of masking is usually a good idea. It seems that absolute
> or proportional thresholds are both quite common, although you could
> also use a gray matter mask, as you suggest.
>
>
>> 3. Most Studies use DARTEL/VBM to investigate Gray matter differences - should it be used to investigate White matter abnormalities?
>
> I don't think there is any principled reason not to do this. I
> suspect that the reason you don't see more VBM for white matter is
> there is that people my feel that DTI or some similar method may
> provide a better measure of tissue integrity (e.g. fractional
> anisotropy) than what you would get with the volume measurements
> provided by VBM. But I imagine there are useful things that can still
> be gleaned from VBM in white matter. Although, perhaps others have
> more experience with this?
>
> Hope this helps!
>
> Best regards,
> Jonathan
>
|