Dear Ruth:
Your methods below bring up several issues.
1) I don't think it is a good idea to threshold your fmap at a
threshold of 0.999 since you will probably include voxels that have
nothing to do with your conditions of interest. This has been debated
before on the list (I don't have references) but I would generally
suggest a threshold of 0.001 uncorrected, although I should add that
this advice is mostly based on the recommendations of others and not
because I have studied different thresholds. The point of my quote
below was to emphasize 2 things: 1) That the contrast does not affect
the data that's extracted (assuming a voxel appeared in different
contrasts)- because SPM goes back to the raw data before adjusting it,
and more importantly 2) that the contrast you choose is a window to
the voxels in that it shows you voxels that are likely to have some
relationship to the conditions you are interested in. Indeed there is
no point in including voxels in your ROI that have no relationship to
the conditions of interest. This is a point that I think Klaas has
made in the past for DCM but applies equally well to PPI.
2) I disagree that one needs to include all voxels in a ROI,
especially for very large ROIs where differences in both an individual
subject's anatomy and function are likely to be influencing the
locations of activated voxels. This is likely to be true even if we
could normalize subjects perfectly, which we cannot do. Therefore
including all voxels in a ROI across your population would almost
certainly mean including voxels whose activity had nothing to do with
your task.
3) The 16 mm radius sphere is very large. I think you would have a
hard time convincing a reviewer (if not yourself) that all the voxels
in such a sphere would really be related to your task. In general
spheres of 4-8 mm radius are used depending on the amount of smoothing
and the location in the brain. I'm not sure what you mean that if you
use a sphere of smaller radius you couldn't find activated voxels in
all your subjects. Are you using a fixed location for the sphere?
Generally I recommend selecting ROI's in a subject specific manner.
One could do this by determining for the group the location of the
activation maxima for a particular contrast and then choosing the
activation for each subject that's "as close as possible" to the xyz
location of the group maxima. (Alternatively the group locus could be
specified based on anatomy or reports in the literature.) This allow
you to select the ROI for each subject so that it is located in the
proper area and contains voxels that are related to your contrast. In
general, depending on smoothing and location in the brain, one would
like the locations of the various subject specific ROI's to be within
1-1.5 cm of each other (or less). If this means that some of your
subjects cannot be included in the PPI analysis, that's what you
should do since those subjects would not appear to have activations
related to the task.
4) I'm not sure I understand how you are extracting the time series
and calculating the PPI. Are you using the eigenvariate button to
extract the time series (in SPM8 or the VOI button in SPM5), followed
by clicking the PPI button which performs the deconvolution and sets
up the PPI interaction term for you? If not how are you deconvolving
the time series? Please note that the first eigenvariate is not a
deconvolved time series. It would be entirely incorrect to form the
PPI term by multiplying the first eigenvariate by the task effects and
reconvolving with an HRF (see Gitelman et al., Neuroimage, 19:200-207,
2003).
Darren
On Mon, Nov 22, 2010 at 6:32 AM, Ruth van Holst <[log in to unmask]> wrote:
> Dear PPI experts,
> I am currently working with Psychophysiological interaction (PPI) analyses
> and have some question about my applied method so far.
>
> I did the following:
>
> To identify the left IFG activation for each participant we examined the F
> contrast of all conditions at p<0.999 uncorrected, to obtain similar amount
> of voxels in or VOI across all subjects. The deconvolved time series from a
> 16 mm radius sphere around the individually defined peak activated voxel
> within the left IFG was extracted (30 participants). The PPI was calculated
> as the element by element product of the left IFG time series (the first
> eigenvariate from all voxels' time series) and a vector coding for the
> effect of task. These products were subsequently re-convolved with the
> hemodynamic response function (HRF). The interaction term was then entered
> as a regressor in a first level model together with the time series of the
> left IFG and the vector coding for the task effect. The models were
> estimated and contrasts generated to test the effects of positive PPI. These
> contrast images for the PPI effects were then entered in a second level
> analysis.
>
>
> These choices were based on he SPM forum list, were I understood the following:
>
> “you should think of the contrast you use to display the activations as
> simply a window to which voxels will be extracted. SPM goes to the xyz
> locations of the active voxels in your VOI, gets the actual data from your
> processed images and then adjusts the data for any effects of interest you
> specify. The contrast you use to display activations merely identifies the
> voxels to extract data from it does not affect the data that are extracted.”
> By Darren Gitelman
>
> Therefore I used a absurdly low threshold of p<0.999 to capture all activity
> in the VOI and to make sure that the below mentioned advise could be
> assured.
>
>
> “Always use ALL the voxels in the ROI. You don't want to compare a subset
> of voxels in one subject to a subset in other subjects, then you might be
> getting apples and oranges. This is of course a bigger issue if you were to
> use a larger ROI.” By Donald McLaren
>
>
>
> I do think that a sphere of 16 radius is quite large, but when using a
> smaller radius I could not detect activity in this region for each
> participant. This would be a problem because excluding participant form this
> analysis would be a waste as well.
>
>
>
> The results from this analyses look sensible, but I am unsure if my method
> is decent enough to pursue writing it down.
>
>
>
> I hope that you are able to give your opinion about my chosen PPI strategy.
>
> Thanks in advance for your time,
>
> Best wishes
>
> Ruth
>
>
>
>
--
Darren Gitelman, MD
710 N. Lake Shore Dr., 1122
Chicago, IL 60611
Ph: (312) 908-8614
Fax: (312) 908-5073
|