Show all posts by user
Dear AFNI users-
We are very pleased to announce that the new AFNI Message Board framework is up! Please join us at:
https://discuss.afni.nimh.nih.gov
Existing user accounts have been migrated, so returning users can login by requesting a password reset. New users can create accounts, as well, through a standard account creation process. Please note that these setup emails might initially go to spam folders (esp. for NIH users!), so please check those locations in the beginning.
The current Message Board discussion threads have been migrated to the new framework. The current Message Board will remain visible, but read-only, for a little while.
Sincerely,
AFNI HQ
History of AFNI updates
Results 2311 - 2340 of 2632
Hi Erin,
> Is it possible, using 3dLME to specify the model so that I can investigate the effect
> of patient status after accounting for any potential effect of scanner (pre- or
> post-upgrade)?
Sure, you can add the scanner as a categorical variable.
> Would I need to change my SS calculation to Type I/sequential and then enter
> scanner first in my model specificatio
by
gang
-
AFNI Message Board
Finally I believe that I've spotted the problem!
You used lower case for 'group' and 'emotion', but when you listed the data table, you somehow decided to switch to 'Group' and 'Emotion'. Once the inconsistency is resolved, hopefully you (and 3dMVM) should be in a happy mood and not at any of the other five levels of the emotion variable!
by
gang
-
AFNI Message Board
Elise, it looks like your AFNI version is a little old. Update AFNI first with the following command:
@update.afni.binaries -d
and then try running the script again.
by
gang
-
AFNI Message Board
Elise,
At the directory where you ran your 3dMVM script, choose one of files that showed up in the error message (that could not be loaded), and run the following command (I'm using stats.251+tlrc'' as an example):
3dinfo -verb stats.251+tlrc''
See what you get from the above command.
by
gang
-
AFNI Message Board
This is weird because I have no trouble running your script here on my computer. I'm running out of ideas except that I may take a closer look if you send me the files plus the 3dMVM script through, for example, Dropbox.
by
gang
-
AFNI Message Board
Elise, could you remove the backslash at the last line and then rerun the script? If you still get the error message, could you paste the whole text from 3dMVM on the terminal here (not just the error message at the end)?
by
gang
-
AFNI Message Board
Elise,
The following two lines in your script need a backslash at the end:
3dMVM
...
Subj Group Emotion InputFile
But I'm not sure whether this is the only problem because that would not give the error message you got.
by
gang
-
AFNI Message Board
How about this: use 3dMean -sd on each group separately to obtain the within-group standard deviation. The ratio of the variances (standard deviation squared) between the two groups would roughly follow an F(n1-1, n2-1)-statistic distribution, which should give you some idea about the variance difference.
by
gang
-
AFNI Message Board
> the results of this test could be biologically meaningful based on previous
> descriptions in the literature. I also want to see if the differences in the
> variance are brain-region specific.
To clarify, you just want to see whether the two groups have the same variability, without the involvement of those three explanatory variables (all within-subject factors?), right?
If so
by
gang
-
AFNI Message Board
> After running my between-subjects GLM (3dMVM), I noticed that the
> variance in one group of subjects appeared to be different than that
> of the other group.
What are the explanatory variables in your 3dMVM analysis in addition to two groups of subjects? And what kind of response variable (input files from each subject) do you have? How did you reach the conclusion that one gr
by
gang
-
AFNI Message Board
> I assume that the -errts would be demeaned as well?
That's correct: The mean is removed through the constant (or intercept) of the polynomials.
by
gang
-
AFNI Message Board
Erik, you need to provide two values for each subject, one for the difference of the two conditions, and the other for the corresponding t-statistic.
by
gang
-
AFNI Message Board
> is there a way to get the residual fMRI time courses which are free of any trends?
Sure. How are you running the preprocessing and analysis, writing your own script or using afni_proc.py/uber_subj.py? You can simply specify the model or modify the script so that you have trend removal only (maybe plus head motion parameters), and the output from option -errts would be what you're lo
by
gang
-
AFNI Message Board
No, there is no way to handle missing modulator values in 3dDeconvolve. Either abandon that explanatory variable, or censor out those few trials (if feasible), or model those few trials as a separate regressor (of no interest).
by
gang
-
AFNI Message Board
Chris,
So you want to see whether the coupling at a few regions under one condition is different under another condition? If the trials vary between 8.5-10 seconds, it's probably not going to make much different the duration difference is less than a TR, but dmBLOCK with -stim_times_AM1 would not hurt either.
I don't have a better suggestion, but maybe you could try:
1) run the
by
gang
-
AFNI Message Board
Rengin, thanks for sharing your experience! The option "-missing_data 0" basically would have impact on those voxels where you don't have any real data, usually outside the brain. So nothing substantial there.
by
gang
-
AFNI Message Board
Elise, it's hard to tell what went wrong without direct access to your data. Two options:
1) Send me the data plus the command lines through Dropbox; or
2) Try out 3dMVM and see how it goes.
GroupAna has been decommissioned for a while since its functionality can be better served by 3dMVM.
by
gang
-
AFNI Message Board
Chris, I have trouble understanding the big picture here. Is this an experiment with three different conditions each of which has a varying duration across trials? Are you trying to calculate the correlations among some pre-selected (and anatomically defined) ROIs under one specific condition (and this is why you want to remove the effects from the other two conditions)?
by
gang
-
AFNI Message Board
I can't see how you could get that error message. Without access to the data, it's hard to pinpoint the culprit. Does the result look reasonable to you?
by
gang
-
AFNI Message Board
Could you explain what those covariates are and provide the content of file covariates_cen.txt?
by
gang
-
AFNI Message Board
Colm, unfortunately the error message from your R 3.0 version is just beyond me. So you can live with 2.15.3 for now?
by
gang
-
AFNI Message Board
Hi Jeff,
> the categorical variables are all between-subjects factors.
In that case you can even use 3dttest++ for the analysis by dummy coding those between-subjects factors as covariates. Plus it does provide the masking option.
> I was hoping for a masking option to speed up my analysis--
> it sounds like the program must be spending most of its run time
> analyzing all
by
gang
-
AFNI Message Board
Hi Jeff,
> I've thus got 1 image for each participant and a mixture of categorical and quantitative predictors.
What kind of categorical variables do you have? Between- or within-subject factors? 3dRegAna can handle between-subjects factors, but it would be quite challenging if you have within-subject factors, and that's why you usually recommend 3dMVM in terms of modeling capa
by
gang
-
AFNI Message Board
Chris,
As far as I know, there is no reasonable approach to fully removing the effects of a specific condition from the FMRI signal. One possible way to do it is to run the full regression model and then use 3dSynthesize to achieve the removal, but the problem is that this can *only* remove the average effect (across all trials) of the condition, and will not take care of the trial-to-trial fl
by
gang
-
AFNI Message Board
OK, my wild guess is that your R version 2.15.3 on your Mac currently points to the 32-bit version, and this was a specific problem for the Mac version of R 2.15.3. If this is the case, two solutions:
1) update your R to 3.0
2) link R to the 64-bit version:
/Library/Frameworks/R.framework/Resources/bin/R64
by
gang
-
AFNI Message Board
> Can anyone tell me please what version fo R is AFNI built against?
I believe the R programs in AFNI are relatively insensitive to the R version. Let me know if you encounter any issues.
by
gang
-
AFNI Message Board
> I am still not sure whether this will add much for me
It's hard to tell because that depends on how much duration variability you have across trials.
> but I am more curious about whether this could be done and if so, would like to do
> it and compare across different modeling options to get a sense of what it is doing.
>
> Is using dmBLOCK for even-related designs
by
gang
-
AFNI Message Board
Emily,
> I would run the ICA script individually for each subject?? Is this a single subject ICA
> approach or GroupICA? ... Additionally, don’t we want to eventually concatenate all
> the subjects? For example I know the GIFT matlab toolbox for ICA , after each
> individuals subjects data are reduced in the data reduction step, the files are
> concatenated into one group a
by
gang
-
AFNI Message Board
Hi Erin,
It looks like that you need to update your AFNI with the following command and re-run the analysis. There was a bug in 3dLME.
@update.afni.binaries -d
> I'm still curious about how to see the direction of the effects...I wasn't able to specify a glt for main effects.
The directionality can be revealed through t-tests. What do you mean when you say that you weren
by
gang
-
AFNI Message Board
Danny,
Does the duration vary only across tasks, or does it change across trials within a task?
> For example,if one trial had a set size of 2 occuring at 16 seconds and with a
> duration of 3.5 seconds, I could specify this by using 16*2:3.5 in my .1D file?
If the duration varies across trials within a task, the modulation approach usually assumes a linear relationship between
by
gang
-
AFNI Message Board