Hi
On Wednesday, Oct 15, 2003, at 03:02 Europe/London, nima dehghani wrote:
> Hi Christian!
> back to you again..and sorry for the delay!
>
> in reply to your reply to my second question:
>
> the best smooth is a filter that is of matched size to the activation
> we wish to detect. The anatomical variability between subjects will
> mean that the signal across subjects may be expected to be rather
> widely distributed over the cortical surface. In such a case it may be
> wiser to use a wide smoothing to detect this signal. In contrast, for
> a single subject experiment, it would be wiser to use a very narrow
> smoothing, or even no smoothing.....
>
> THIS IS FOR GLM ANALYSIS!..THE SAME LOGIC COULD ALSO BE ASCRIBED TO
> ICA (i think so,...maybe you do not agree with me)
>
> if we are using ICA:
> & if we are analyzing data of ONE subject, there is no prob; as we can
> do it with least gaussian (or even no guassian..as you pointed)
> smoothing!
> but if we need data across subjects; and we do the same, we will be in
> danger of lossing significant points!
> what is your idea?
>
Again, the optimal procedure will depend on what you propose to do for
a multi-subject study; both w.r.t. to width of a smoothing kernel and
w.r.t. the question of individual vs. unique smoothing kernel.
Let's assume for the moment that you would like to combine IC maps from
individual subjects. Then at the individual subject level the optimal
filter should match the expected activation size. Ideally, one could
employ some form of non-linear filter in order to preserve intensity
edges and to avoid 'blurring' across boundaries. As the 'true'
activation size is unknown it makes perfectly sense to use the same
smoothing width for each subject and match the width to one single
expected activation size. If you now want to combine individual maps at
some kind of group level, registration issues come into play. At this
point the 'optimal' thing to do is to get a really good registration in
order to avoid any kind of additional blurring. If registration is not
perfect, some smoothing could be used where now the smoothing kernel
width should best match the expected displacement of an individual
voxel across subjects.
> and... what are the practical steps for what you called
> "concatenating"?
>
>
the command line tool 'avwmerge' allows you to concatenate data in
x,y,z or t.
ta
christian
|