Again, Steve, thanks for your help.
I need some more clarification on few other things:
>Hi,
>
>On 28 Oct 2006, at 21:31, Michael wrote:
>
>>> Next, the most likely difference is that fstats don't distinguish
>>> between negative and postitive activation - both show up as
>>> "positive" - this is the nature of fstats. Hence your zstats only
>>> show positive, the fstats show both.
>>
>> So my fstats setup was correct then?
>
>Yes - if you want to ask the question "where does any linear
>combination of the 3 EVs produce a significant response?"
>
Okay, so the results I get from the setup are 3 different zfstat maps:
zfstat map or EV1
zfstat map for EV2
zfstat map for EV3
I always assumed that each zfstat map explained the variance for the
respective contrast.
For example:
zfstat actavition map1 is the activated ROI's explained by only EV1.
zfstat actavition map2 is the activated ROI's explained by only EV2.
zfstat actavition map3 is the activated ROI's explained by only EV3.
It seems that way, when you look at each activation map, because you see
different areas of the brain activated for the respective zfstat map.
Is the correct?
>> And why does the contrast setup:
>>
>>>> EV1 EV2 EV3 F1 F2 F3
>>>> OC1 * 1 0 0 * 0 0
>>>> OC2 * 0 1 0 0 * 0
>>>> OC3 * 0 0 1 0 0 *
>>
>> only represent or detect positive activation ONLY as opposed to the
>> F-test
>> which detects both positive and negative activation? I would think
>> they
>> would do the same thing, no?
>
>No - it just comes down to the definition of the fstat - it's
>basically asking how much variance is explained by the space spanned
>by the EVs included in the fstat - kind of like a sum-of-squares of
>the individual effects - it is always positive. Whereas simple
>contrasts, generating tstats, are always signed.
>
I guess what I'm asking is that in the past with other statistical
packages, I would see the same results with the F-stat and t-stat maps.
The only difference being, obviously, the sign and which statistic it
showed, F or t, but the same areas of activation and number of voxels were
presenct for both F-stat map and t-tstat map.
Say if you ran a paired t-test and ran a repeated measures ANOVA with 2
levels, wouldn't you get the same results basically despite one having t-
stat for a results and the other has the F-stat?
But I guess in the case w/ FSL, when setting up the contrast, it will
always just show results for positive activation? Is there a way to
indicate wanting to detect both positive and negative activation in the
contrast? Or do we always have to include the F-stat for that?
>Note that you can always restrict the fstat reporting to areas where
>one or more of the individual contrasts are positive, using the post-
>contrast masking button in FEAT.
>
>Cheers.
>
>------------------------------------------------------------------------
>---
>Stephen M. Smith, Professor of Biomedical Engineering
>Associate Director, Oxford University FMRIB Centre
>
>FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>+44 (0) 1865 222726 (fax 222717)
>[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>------------------------------------------------------------------------
>---
>=========================================================================
|