| T O P I C R E V I E W |
| dvsmith |
Posted - Mar 02 2011 : 7:03:31 PM
I'm trying to use one of the FSL tools called unconfound. Its usage is simple, yet it keeps throwing errors when I try it:
[smith@node53 run1.ica]$ unconfound
Usage: unconfound <in4d> <out4d> <confound.mat>
[smith@node53 run1.ica]$ pwd
/home/smith/experiments/Imagene.02/Analysis/TaskData/10156/Framing/MELODIC_150/Smooth_6mm/run1.ica
[smith@node53 run1.ica]$ unconfound filtered_func_data.nii.gz denoised_tmp.nii.gz for_confound.txt
terminate called after throwing an instance of 'RBD_COMMON::BaseException'
Aborted
I've tried change the extension from .txt to .mat, and I've also tried running dos2unix to no avail. I suspect this is an FSL bug, but I wanted to make sure someone else was also seeing the error before I write them.
Thanks!
David
|
| 4 L A T E S T R E P L I E S (Newest First) |
| dvsmith |
Posted - Mar 04 2011 : 1:40:55 PM
Just to follow up with this... Apparently, the unconfound.mat file needs to be in the exact same format as a design.mat file that Feat would use. I think an easy workaround would be to put your confound txt file into the confound EVs options in Feat and then generate the design.mat file using feat_model. The only catch here (I think) is that you must trick Feat into thinking it has at least one "real" EV in its design matrix, so just make a design with a single empty EV. I'm waiting for someone at FMRIB to confirm that my approach is OK: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;6856cd8f.1103
|
| dvsmith |
Posted - Mar 03 2011 : 12:54:44 PM
ok -- thanks, Chris. Maybe it's a new bug since nobody complained about it in older posts. I'll email the list to see what's up. |
| petty |
Posted - Mar 03 2011 : 12:40:20 PM
I tested a couple things last night, suspecting it could be the version of fsl and your particular datatype.
It was reproducible in the last 2 releases of fsl, and it threw the same error on float32 and int16 data.
|
| dvsmith |
Posted - Mar 03 2011 : 12:21:16 PM
I can reproduce this error on three different systems: OSX (10.6.6), Ubuntu 10.04, and the cluster. I've tried the latest 3 versions of FSL. As far as I can tell, there isn't anything wrong/corrupt with the inputs I'm using, but I could be wrong here... Nobody else has complained of this bug on the FSL forums, so I'm assuming I'm just doing something wrong... |
