| Author |
 Topic  |
|
|
crozier
New Member
19 Posts |
 Posted - Jun 13 2007 : 10:02:35 AM
|
Hi,
If I run FEATQUERY with a 27-voxel mask on a higher-level analysis (across subjects) I get a file called tsplotc_zstat1.txt which contains the data for each of the individuals contributing to the higher level analysis. According to the text, the values in column 1 should be the average parameter estimate across the 27 voxels for each individual (is this correct?). However, when I run FEATQUERY on the second-level analyses (runs within an individual) the average parameter estimate for the 27 voxels is similar but not exact.
Does anyone know why these two wouldn't be the same?
Joe
|
|
|
crozier
New Member
19 Posts |
Posted - Jun 13 2007 : 10:40:42 AM
|
I've pasted below an exchange from the archives of the FSL email list. It looks like for tsplotc the voxels within the mask are weighted by the z-scores. Can I turn this off when I call FEATQUERY with the -n option or does that refer to how the program is configured on the server?
"Hi - good question. The reason is that by default, tsplot produces these
mean timecourses from the data using weighting by the zstat images (within
the mask chosen). Hence the difference contrasts get different mean
timeseries as they have different weighting functions.
If you want to change this (ie turn off this within-mask weighting), edit
$FSLDIR/bin/featquery and change the call to tsplot to include the "-n"
optoin.
Cheers, Steve.
On Sun, 12 Jun 2005, Henry Lütcke wrote:
> Dear FSL Experts,
>
> I have performed featquery in order to get the signal time course
> averaged across all voxels in a region-of-interest. I guess this should
> be the first column in any tsplotc_zstat*.txt file in the
> featquery\tsplot directory. I find different values in the first column
> of the respective file for each of the five contrasts I have performed.
> Should the mean signal time course not be the same for all contrasts,
> since I am looking at the data for all voxels in a ROI?
>
> Thanks a lot for your help.
>
> Henry Lütcke
>
>
>
>
|
 |
|
|
syam.gadde
BIAC Staff
USA
421 Posts |
Posted - Jun 13 2007 : 10:45:46 AM
|
| They are suggesting editing the featquery script. As you can't edit the installed version, you could copy it and edit it yourself, and run the edited copy. Make sure to use nedit or some golgi-based editor to edit it (rather than Wordpad, etc.) |
|
|
| |
Topic |
|
|
|
Question about tsplot file from FEATQUERY
