Dear Dana,
Indeed doing a SVC on a single voxel point makes no sense. In my opinion
(and if I understand your question correctly) the thing to do is to use the
specific brain region as you hypothesized a priori. If you know is the same
area of activation, it is not necessary to use the specific coordinates from
the literature because they might show little differences from your
coordinates due to differences in data collection (PET camera, data
reconstruction) or data analyses (SPM version, normalisation parameters,
smoothing kernel, etc).
It is oke to use the uncorrected p-values to detect a priori hypothesized
regions. But be aware of the size and shape of the volume you specify for
SVC. To my knowledge: size and shape of your small volume must be as close
as possible to (certainly not smaller than) the size and shape of your a
priori specific brain region ("you should define the boundaries of your
region a priori" see below).
Regards,
Simone.
On Wed, 2 Oct 2002, Grant, Steven (NIDA) wrote:
> Doing a SVC on a single voxel point makes no sense. My understanding is
> that you should define the boundaries of your region a priori, then perform
> the SVC irrespective of the outcome of the whole brain analysis.
>
> sg
>
> -----Original Message-----
> From: Dana Eldreth [mailto:[log in to unmask]]
> Sent: Wednesday, October 02, 2002 3:59 PM
> To: [log in to unmask]
> Subject: Small Volume Correction in PET O15
>
>
> Hello everyone,
> I have a question concerning when you can small volume correct and when
> you can't small volume correct (PET O15)
>
> If you hypothesize a priori that a specific brain region will show
> activation during a specific task and this region shows activation after
> you set the height and extent thresholds for your results (uncorrected),
> can you SVC this voxel point since you hypothesized this region a
> priori?
>
> Or... must you have a specific coordinates from the literature using a
> similar task?
>
> If anyone could please help clarify this issue, it would make my day.
>
> Thanks,
> Dana
>
|