Dear all,
I've spent a bit of time trying to specify a model that I want to run with 3dLME and ran across a few things:
I've got a simple model (2 subjects, 2 factors: task with 2 levels and freq with 3 levels) with 6 .nii files associated with these. The problem is that I'm not quite sure wether an explicit nii file can be associated, regardsless the filename or wheter each of the file needs to be named according a certain scheme.
For example, when I run the 3dLME command with the following simple model on some artificial data (for now, just to get a proof-of-principle before I'm going to include more factors, levels and random effects [ranEff]):
3dLME \
-jobs 4 \
-prefix tmp \
-mask mask.nii \
-model "task*freq" \
-SS_type 3 \
-dataTable \
Subj task freq InputFile \
s1 not 250 s1/not/beta_0001.nii \
s1 not 750 s1/not/beta_0002.nii \
s1 not 3000 s1/not/beta_0003.nii \
s1 aud 250 s1/aud/beta_0001.nii \
s1 aud 750 s1/aud/beta_0002.nii \
s1 aud 3000 s1/aud/beta_0003.nii \
s2 not 250 s2/not/beta_0001.nii \
s2 not 750 s2/not/beta_0002.nii \
s2 not 3000 s2/not/beta_0003.nii \
s2 aud 250 s2/aud/beta_0001.nii \
s2 aud 750 s2/aud/beta_0002.nii \
s2 aud 3000 s2/aud/beta_0003.nii \
I get some output that makes me think the name of the file should be some kind of contraction of the factors and levels:
++++++++++++++++++++++++++++++++++++++++++++++++++++
***** Summary information of data structure *****
3 subjects : s1/aud/beta_00750s1/aud/beta_0002.nii s1/not/beta_00250s1/not/beta_0001.nii s2/not/beta_003000s2/not/beta_0003.nii
3 response values
3 levels for factor task : s1/aud/beta_003000s1/aud/beta_0003.nii s1/not/beta_00750s1/not/beta_0002.nii s2/aud/beta_00250s2/aud/beta_0001.nii
3 levels for factor freq : s1/not/beta_003000s1/not/beta_0003.nii s2/aud/beta_00750s2/aud/beta_0002.nii s2/not/beta_00250s2/not/beta_0001.nii
0 post hoc tests
For example, I do not have 3 subjects (only 2); and the data is structure seems wrong across the factors (e.g., I don't have 3 levels for task). Then it continuous with "Tabulation of subjects against all categorical variables", again with garbled data. Moreover, it then crashes when trying to load the data:
Reading input files now...
** ERROR: Dset s1/aud/beta_00250s1/aud/beta_0001.nii could not be loaded
** ERROR: Dset s1/aud/beta_00250s1/aud/beta_0001.nii could not be loaded
** ERROR: Dset s2/not/beta_00750s2/not/beta_0002.nii could not be loaded
** ERROR: Dset s2/aud/beta_003000s2/aud/beta_0003.nii could not be loaded
Error in dim(inData) <- c(dimx, dimy, dimz, NoFile) :
This makes me think that each of the filenames should be a logical combination of subject(name), factor and level, rather than just a reference to an existing file (e.g. SPM or FSL output). Is this assumption correct or is something else at miss...
Cheers,
Cris