Hi Darren,
On 14 Apr 2006, at 09:55, Darren Schreiber wrote:
> Mechanically, it looks like featquery worked great. I downloaded
> the second level FacesContrast11.gfeat directory from the cluster
> where I did the analysis. Then, I chose the cope1.feat directory
> and asked featquery to give me all the available stats (stat/pe 1 2
> 3, stats/cope, stats/varcope, stats/tstat, stats/zstat,
> thresh_zstat). I selected a mask file. Asked it to "Convert PE/
> COPE values to %" and hit "Go."
>
> This was the text output:
> /usr/local/fsl/bin/featquery 1 /Users/dschreib/Documents/
> FaceContrast11.gfeat/FaceContrast11.gfeat/cope1.feat 8 stats/pe1
> stats/pe2 stats/pe3 stats/cope1 stats/varcope1 stats/tstat1 stats/
> zstat1 thresh_zstat1 featquery -p -s -b /Users/dschreib/Documents/
> FaceContrast11.gfeat/FaceContrast11.gfeat/ROIs/LeftAmygdala_0000.hdr
>
> However, I am not sure how to interpret the results reported on the
> webpage.
>
> With other software that I have used to analyze ROIs, I have
> calculated a % signal change from a resting baseline. Here, I am
> extracting ROIs from a second level contrast that I ran where
> activations while subjects were looking at black faces were
> contrasted against activations while they looking at white faces.
> I think that the stat/tstat1, stat/zstat1, and thresh_zstat1
> probably are not meaningful in this context, right? But, I am not
> sure whether pe1, pe2, pe3, cope1, or varcope1 contain data that is
> useful for me either.
>
> Part of my problem is that I still don't have a good intuition for
> pe's, copes, and varcopes. Pe's are equivalent to beta's right?
> And, a cope is contrasting one pe with another?
That's right. And varcope is the estimated variance of the cope.
> The reported mean value of cope1 is 300.04, so does that mean that
> amygdala activation is 300% greater for seeing these black faces
> than white faces?
If you have used the latest version of FEAT for first- and higher-
level analyses then yes you could trivially interpret the featquery "%
cope" values at first- and higher-level. However you probably have an
older analysis here. If your first-level regressors were of height 1
and you had simple contrasts and you also had a simple model
+contrasts at second-level (e.g. just a group mean) then probably the
easiest thing to do is to work things out simply: e.g., if you turn
off the "%" button in featquery, then your cope values reported from
your gfeat are:
COPE as a % signal change = second-level-cope (meaned over subjects)
* 100 / 10000 (because the signal at first level was probably scaled
to a mean of about 10000).
If you want to get more accurate than this (e.g. if you have a more
complex second-level design), the simplest thing to do is to just re-
run FEAT from scratch.....
Cheers, Steve.
> I've gone through the new online course materials (they are great!)
> and looked around through the other FSL pages. But, I haven't
> found a good example for utilizing and interpreting the data
> extracted with featquery. And, any references on pes, copes, and
> varcopes would help. I can tell my intuitions for these are poor.
>
> I really appreciate all the help and the work that's been going
> into making FSL even better.
>
> Darren
>
> On Apr 12, 2006, at 9:22 PM, Steve Smith wrote:
>
>> Hi - should be straightforward. If you are wanting to look at PEs
>> then featquery looks in design.mat for the line containing
>> PPheights and if it's COPEs then it looks in design.con for the
>> same thing. This determines, for PEs or COPEs, what the peak-peak
>> height of the design matrix is. If it doesn't find it in your
>> files it assumes the height is 1. So you can check for that, and
>> if it's not set in your files, and you're worried about the peak-
>> peak height, you can always set that by hand to avoid rerunning
>> the analysis. and example file is attached.
>>
>> Cheers, Steve.
>>
>> <design.con>
>>
>>
>>
>> On 13 Apr 2006, at 05:11, Darren Schreiber wrote:
>>
>>> I used feat version 5.0 to analyze some data on a cluster a while
>>> back. I see that featquery now has a GUI version that will
>>> extract ROIs and calculate percent changes in signal.
>>>
>>> Is it possible to use the current version of featquery to look at
>>> ROIs on data that I analyzed with the early version of feat? Are
>>> there any problems that I should anticipate with it?
>>>
>>> I'm looking forward to using the new version of FSL with my new
>>> data, but I don't want to completely rerun all of my analyses on
>>> my old data if possible.
>>>
>>> Darren
>>>
>>> ********************************************************************
>>> ***********
>>> Darren Schreiber, J.D.
>>> Assistant Professor
>>> Political Science, SSB 367
>>> 9500 Gilman Drive
>>> La Jolla, CA 92093-0521
>>> dmschreiber (at) ucsd (dot) edu
>>> ********************************************************************
>>> ***********
>>
>>
>> ---------------------------------------------------------------------
>> ------
>> 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
>> ---------------------------------------------------------------------
>> ------
>>
>>
>>
------------------------------------------------------------------------
---
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
------------------------------------------------------------------------
---
|
