| T O P I C R E V I E W |
| mullette-gillman |
Posted - Sep 29 2007 : 12:21:08 PM
In my FSL analysis I am including a nuisance EV for missed trials. In some runs subjects do not miss a single trial, so there are no instances of this event. In FSL 3 I included a 3-column file with just a single row of 3 zeros. When I entered in the design in the GUI I received a warning that: "Problem with procesing the model: Warning: at least one EV is (close to) a linear combination of the others. You probably need to alter your design. ...", but the analyses would run.
In FSL 4, I receive the same warning, and when I try to run the model I receive a new window that says: "Please setup model before running"
Has anyone else experienced this? Any advice on how to deal with thi in FSL 4?
|
| 10 L A T E S T R E P L I E S (Newest First) |
| dvsmith |
Posted - Aug 08 2009 : 10:09:11 AM
Make a new template and use the option that looks for "Inputs are 3D cope images from FEAT directories" -- i.e., <whatever>.feat/stats/cope<number>.nii.gz
You should only select the valid cope images for each subject (i.e., those that are not empty). You'll have to do this for each cope just like a 3rd level analysis.
David
|
| Carpenter |
Posted - Aug 07 2009 : 3:55:01 PM
Hi,
I've been having problems with 3rd level analyses stalling when I use 2nd levels where some runs had empty EVs. I came across this thread (as well as several on fsl's website) where it appears people are running separate analyses for each condition to avoid carrying up the empty EVs, however I cannot figure out how to set up my 2nd level template to do it this way. I hope that makes sense. Any ideas? Thanks!
Kim |
| wang |
Posted - Feb 19 2008 : 1:36:39 PM
Oh, I used "inputs are low-level FEAT directories". By changing it to "Inputs are 3D cope images from FEAT directories" it works now. Thanks a lot Chris! |
| petty |
Posted - Feb 19 2008 : 1:26:11 PM
are you selecting:"Inputs are 3D cope images from FEAT directories" |
| wang |
Posted - Feb 19 2008 : 12:53:11 PM
I entered 0 0 0 in fsl 3-column file for the empty EV. It was okay for the first level analysis. The output of the empty EV has an empty image. But I am having trouble in the second level analysis. I'm trying to do the second level analysis cope-by-cope, so that all non-empty EVs can be included into the analysis, somehow fsl does not run this one cope each time model.
Has anyone had success in doing this? Thanks. |
| petty |
Posted - Nov 14 2007 : 1:30:33 PM
quote:
Originally posted by mullette-gillman
If you hear about the patch coming out, could you let me know?
4.0.2 is now available: http://fsl.fmrib.ox.ac.uk/fsldownloads/patches/
quote:
FSL people
FSL 4.0.2 is now available:
Susan multiple performance improvements
SienaX Bugfix in sienax regarding the -lm option Conf/doc minor changes IRTK linux 32/64 flavours Avwutils minor fixes to fslmaths and fslsplit
Feat5 Multiple minor fixes
Film Minor fix
Flame bugfix in the multiple-group-variances estimation Fdt minor changes and fixes Miscvis Fixed Renderstats gui
|
| mullette-gillman |
Posted - Oct 11 2007 : 11:17:18 AM
This specific EV is for missed trials. I include it only to remove a rare nuisance variable. I am not running any contrasts against it.
I ran it with '0 0 0' for the 3-column file.
If you hear about the patch coming out, could you let me know?
Thanks!
O |
| petty |
Posted - Oct 11 2007 : 11:08:44 AM
did you run it with three zeros in the batch file, or with empty EV selected?
If you ran it as '0 0 0', DONOT include this subject's EV into your group level, or any of the contrasts with a real EV vs empty ... you won't get any activation.
also, FSL recognized the problems with the empty ev and error messages and they are supposed to fix that in an upcoming patch (they said possibly next week) |
| mullette-gillman |
Posted - Oct 11 2007 : 10:36:33 AM
Thanks!
I actually kept getting the error when I did that. I found a work-around by analysing the run in a batch script. The error still presents in the log file, but the analyses are performed. |
| petty |
Posted - Oct 01 2007 : 09:30:05 AM
in FSL 4 you can either leave your ev "empty" ... under convolution, select empty instead of an hdr shape. |
