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 2461 - 2490 of 2632
Becky, it looks like you have some extra spaces after each backslash (\) in your script.
Also you need to set your model with option '-model'!
by
gang
-
AFNI Message Board
> I'm wondering the extent to which this boot amp will cover some of the newer
> (more advanced) AFNI features added over the past few years such as 3dLME,
> 3dREMLfit, granger causality, DTI, 3dMVM, 3dMEMA, ETC.
Hi Adam, we usually cover those topics/programs during the AFNI workshop except for DTI. However, you can still talk to Daniel Glen during the workshop about the pro
by
gang
-
AFNI Message Board
Each of your NIfTI file contains multiple 3D data? Could you paste your script?
by
gang
-
AFNI Message Board
Cindy,
It's hard to tell what's going on with no access to your data. You may consider using 3dMVM instead:
by
gang
-
AFNI Message Board
> It looks to me like the installation chokes during an attempted compiling
> process of ‘mvtnorm’ using a gfortran compiler that is not found. Is this
> the case or have I messed something up in the install process?
Yes, you need to install the gfortran compiler first before installing the lme4 package.
by
gang
-
AFNI Message Board
Yes, you need to install R and three R packages: afex, phia and snow. See here:
by
gang
-
AFNI Message Board
Could you provide the output of the following three commands?
which afni
which R
which 3dMVM
by
gang
-
AFNI Message Board
Please be precise: do you have both 3dMVM and 3dMVM.R in your binary directory? Are you able to open 3dMVM too? And did you open a new terminal and type '3dMVM -help | less'?
by
gang
-
AFNI Message Board
Becky, did you open a new terminal when you typed "3dMVM -help | less"? Also check if you have the following files in your AFNI binary directory: 3dMVM and 3dMVM.R
by
gang
-
AFNI Message Board
It's hard to tell what's going on without access to the data. On the other hand, the analysis can be easily done with 3dANOVA2 -type 3.
by
gang
-
AFNI Message Board
> the t-statistic of this contrast, and the subblock right before it:
> Face-House#0_Coef. I wonder if this sublock is exactly
> "b_face - b_house". What does the #0 stand for?
Yes, that's the effect estimate for the contrast of the two conditions. The pound sign (#) is the index indicator that shows the order of effect estimates (i.e., #0, #1, #2, ...) when multip
by
gang
-
AFNI Message Board
Jackie,
Without access to your data, it's hard to tell what went wrong. How did you define your ROI? And what script did you write to extract the ROI time series?
by
gang
-
AFNI Message Board
3dMVM is now available in the most recent version of AFNI. You can update it with
@update.afni.binaries -d
You may need to open a new terminal to see it.
by
gang
-
AFNI Message Board
> We have 14 subjects. So that would be df = 11, correct?
> 14 - 2(variables) - 1. I think this is also verified in the 3dinfo
> output of the results subrick for the t-stat (located under,
> statcode = fitt; statpar = 11) Correct?
That's right.
For future reference, I've created a web page for the conversion between t-statistic and correlation coefficient:
by
gang
-
AFNI Message Board
> is there a round about way to calculate the partial R^2 with 3dcalc?
I just did a back-of-the-envelope calculation. You can obtain R^2 for an explanatory variable using the following formula with 3dcalc:
R^2 = t^2 / (t^2 + DF)
where DF is the degrees of freedom for the t-statistic.
by
gang
-
AFNI Message Board
> ust to clarify, the parameter estimate B1 for variable x1(Age) is the effect of
> Age on y,after controlling for variable x2 (IQ), correct?
Yes, correct.
> And the parameter estimate B2 for variable x2 (IQ) is the effect of IQ on y
> after controlling for the variable x1 (age), correct?
Right.
> I tried using -rcoef for each regression coefficient in the model. H
by
gang
-
AFNI Message Board
> Is there a way to reference a certain row within the timing file?
I'm not so sure, byt you may try if it works by using something like
YourTimingFile.1D'{INDEX}'
The INDEX starts from 0 as usual. The square brackets [] are for columns.
by
gang
-
AFNI Message Board
Hi Elisa,
3dMVM should work for your case, and it should be available in the upcoming new AFNI distribution (in a couple of days, I hope).
See here:
by
gang
-
AFNI Message Board
A few comments:
1) From an investigator's perspective, you may consider an explanatory variable as being of interest or of no interest; however, from modeling perspective an explanatory variable is treated not differentially regardless how the user views it. In other words, once you put an explanatory variable in the model, it will account for some data variability. In the context of your
by
gang
-
AFNI Message Board
This is what I said initially:
"Alternatively you can analyze each subject separately, and then run fixed-effects analysis at group level."
The weighted least squares is performed (with 3dcalc for the formula on that slide) after the separate individual subject analysis.
by
gang
-
AFNI Message Board
> I could add them manually, but an automated way would be nice :).
Brain exercise should be good for cognitive maintenance.
> I see the methods on page 36; that might make things a bit easier.
> How would one go about implementing that though?
A one line script with 3dcalc should work. Let me know if you have trouble getting it done.
by
gang
-
AFNI Message Board
Maybe I'm not fully understanding your issue, but would it involve too much work to manually add asterisks in those timing files?
Alternatively you can analyze each subject separately, and then run fixed-effects analysis at group level. See methods mentioned on page 36 of the following AFNI workshop slides:
by
gang
-
AFNI Message Board
What are your 'independent variable' and 'covariate'? Could you be a little more specific? In general, 3dttest++ and 3dMEMA are better choices than 3dRegAna.
by
gang
-
AFNI Message Board
> What do you mean by "check the significance of polynomial coefficients"?
As I mentioned last time, if you add option -bout in your 3dDeconvolve script, you would obtain the regression coefficients and their t-statistics for those polynomials fitting, and you can check their significance then.
> When running 3dDeconvolve, I don't specify the polynomial order (I use &
by
gang
-
AFNI Message Board
3dttest++, 3dMEMA, and 3dRegAna all should work for your case. Be careful about the centering issue when dealing with a covariate. 3dttest++ and 3dMEMA have an option for centering, while with 3dRegAna the user has to center the covaraite first before feeding the covariate values into the program.
by
gang
-
AFNI Message Board
How many groups of subjects do you have? Any within-subject factors?
by
gang
-
AFNI Message Board
> Is the response magnitude equivalent to % signal change?
If the EPI signal was properly handled (e.g., scaled by the average) during pre-processing, yes the response magnitude can be interpreted as percent signal change.
by
gang
-
AFNI Message Board
> I want to extract the mean activation from predefined ROIs for each individual.
What do you mean by 'mean activation'? Response magnitude under a specific condition? If so, it would be a regression coefficient or the contrast between two conditions. Also if the predefined ROI is in standard space, make sure that the input file for the extraction is in the same standard space.
by
gang
-
AFNI Message Board
3dMVM should be available in the next build, most likely in a couple of days.
by
gang
-
AFNI Message Board
Hi Chris,
If your stimulus duration is within the range of 8.5 -10.5 seconds, the impulse response shape would probably get lost due to the convolution, and dmBLOCK is not necessarily a bad choice. If you really care about response shape, you may try -stim_times_AM2 combined with TENT. You may have to try out different approaches and find out a reasonable solution. It's hard to see which
by
gang
-
AFNI Message Board