Hello Nicky
The most obvious difference between SPM SVC and a WFU_PickAtlas ROI is
that the glass brain show all activations in SVC while ROI shows only
those activations in the masked area. See attached image. Left is SVC
and right is ROI. Additionally there are smaller differences in the
statistics as the SVC method and ROI methods differ slightly. Mainly
that the mask is resliced to that of the con image in the ROI analysis.
The SVC would be the same if the mask image was also resized.
There is an update for the WFU_PickAtlas 2.4 for SPM8 found at:
http://www.nitrc.org/frs/?group_id=46. Be sure to get the one labeled:
PickAtlas_SPM8_Update_20091123.zip. Alternatively you could upgrade to
the PickAtlas 3.0 (SPM8 compatible only). That link is also available
there.
Please let me know if there are any other questions.
Ben Wagner
ANSIR Lab
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]]
On Behalf Of <Nicky> <Kochan>
Sent: Friday, August 13, 2010 2:32 AM
To: [log in to unmask]
Subject: [SPM] ROI vs SVC
Dear SPMers
I am using SPM 5 and have selected the ROI button to perform a
t-contrast in an a priori region (bilateral hippocampus) defined using
PickAtlas. When reporting this step, I called it a region-of-interest
analysis however a reviewer has pointed out that this is actually a
small volume correction. I note that in the Maldjian paper, it is stated
that SPM will perform a small volume correction when this step is
selected. I also reported a whole brain analysis in addition to this ROI
or SVC analysis.
A couple of questions:
How is a region-of-interest analysis different to this?
Can this same procedure be done in SPM 8 as there is no ROI button?
The reviewer has also asked that I project the significant cluster in
the hippocampus onto a slice view allowing surrounding areas that
survive the same "relaxed" threshold to be seen. I am not sure how I can
do this since the cluster was observed using the ROI button which
selects only the region specified and not surrownding regions. I do not
know how I can estimate an appropriate threshold at the whole brain
level which would allow the hippocampal cluster and surrounding
supratheshold areas to be shown. Any ideas?
Many thanks for your advice,
Regards
Nicky Kochan
|