Dear all,
Could somebody advise on the relative merits of using the existing SVC
tool, versus making your small search region explicit during the
statistical modelling part of SPM?
When you use the SVC tool and select a region, you get corrected p
values only for those voxels which fall within the region *and* were
intially activated at whatever threshold was used for whole-brain
correction. So far in a quick poll of some users I've had someone
suggest an initial threshold of p<0.001 uncorrected, and someone say
they set the p value to close to 1, in order to force more voxels to
be included.
If you specify your small region at the explicit mask stage of the
statistical model, you are forcing all the voxels to be included at
that stage, and can then threshold it accordingly. This also allows
you to see a t-map of only the region you're focused on. To my mind
both these things seem advantageous.
I did a quick experiment with firstly a whole-brain threshold of 0.99,
and then applying a small volume correction, vs. specifying the small
volume as an explicit mask in my model. The former method included
121424 voxels (93 resels), the latter 122673 voxels (56.6 resels).
Peaks were similar but not the same, and there were more peaks using
the SVC tool than by specifying the small volume explicitly.
There is a previous post on this but it only seems to cover whether
it's ok to use the explicit mask option in order to visualise the t
map for the small search volume.
I'd be very grateful for any advice on this; as I said above, the
explicit mask option seems intuitively to do what I'd expect SVC to do
(looks at all the voxels in the region, and shows you the appropriate
t map). Why does the SVC button only look at voxels which already
exceed an (arbitrary) threshold? Is there a recommended whole-brain
threshold to apply before using the SVC button, and/or is it ok to set
this as near to 1 as possible?
Thanks very much,
Susie
|