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 2221 - 2250 of 2632
Do you have multiple values per subject under each condition (level of factor A)? If yes, you use option -fb or -fab. With -fab, you will get two sub-bricks with the first being MSB and the second MSE. Then you obtain MSE using 3dcalc.
However, if the answer is no, then there is no way to differentiate MSAB and MSE, and you would simply consider MSAB as the within-subject MSE.
by
gang
-
AFNI Message Board
> The average cost value is significantly different between groups. So this means I need to add
> -center DIFF to my 3dttest++ script correct? How does this change the interpretation?
There are usually two research goals with a situation yours:
1) examine and compare the two groups while controlling for the covariate;
2) examine the correlation between BOLD response and the covariat
by
gang
-
AFNI Message Board
> Is there a command in AFNI to calculate the MS_between, MS_within, MSE etc. from a
> Repeated Measures design (3dANOVA2 -type 3)?
3dANOVA2 -type 3 is for one-way repeated-measures (or within-subject) ANOVA. Using option -fa you will have two sub-bricks: the first one is sqrt(MSA), the square root of MS (mean squares) for the repeated-measures factor (A), while the second one is the a
by
gang
-
AFNI Message Board
The first question is, do the two groups have roughly the same average cost value? Or do they differ significantly? This question will have a big impact on your analysis result and interpretation. You may read the discussion about the centering issue in the 3dttest++ -help. Once this is clarified, we can talk about the result interpretation.
by
gang
-
AFNI Message Board
> Is the first basis function of the SPMG2 analysis (sub-brick[0]) the activation
> removing the effect of the reaction time (becuase reaction time is run as a
> second regressor) or is the first basis sub-brick showing the activation regardless
> of the effect from other regressors like reaction time.
Using your result as an example:
# -- At sub-brick #66 'dual_hit#0_C
by
gang
-
AFNI Message Board
It's difficult to interpret the results without knowing how your model looks like. Could you provide your model or script? Also what is the output of the following command?
3dinfo -verb outputFrom3dttest++
by
gang
-
AFNI Message Board
> For the group analysis, I want to look at reaction time as a regressor on a trial by trial basis
> between the young and old group. In each trial, the duration of the stimulus and trial was
> held constant. I think I only want to model the BOLD amplitude based on this design, correct?
Yes most people use the effect associated with the first basis function in SPMG2 for group analys
by
gang
-
AFNI Message Board
No, I never saw such duplicate display before. If you upload your data and the 3dMVM script, I can take a close look at the situation.
by
gang
-
AFNI Message Board
Hi Sarah,
The duplicate display of some z-slices is strange, but the runtime seems normal. I would just let it finish and check the results.
by
gang
-
AFNI Message Board
> I want to model the amplitude of the BOLD response, but not the duration.
Could you clarify a little bit? It's not clear to me exactly what your goal is at group level. If you just want to model the BOLD response amplitude, one possibility is to ignore the derivative part and run group analysis only one the effect from the 1st basis function. And this is what most people do.
by
gang
-
AFNI Message Board
Alternatively you can modify the script a little bit and use 3dMVM instead, and see it would go through those linear tests.
by
gang
-
AFNI Message Board
> Does 'group' need to specified for these?
No, not really.
The difficulty is that your original script works just fine on my machine. Without access to your computer, I don't know what to say...
by
gang
-
AFNI Message Board
Hi Keely,
> Are dual_hit#0 and dual_hit#1 the first regressor (intensity) and its derivative?
> (and dual_hit#2 is the reaction time regressor and dual_hit#3 its derivative?)
Yes I believe your interpretation is correct.
> -expr 'sqrt((a^2)+(b^2))' \
Using that value for group analysis could be problematic, and I think there is some discussion about its limitatio
by
gang
-
AFNI Message Board
What's the error message? Same as before?
Tabs seem to have caused trouble before, so using spaces is definitely safer and preferred.
by
gang
-
AFNI Message Board
> can I do a Session X Condition interaction within 3dDeconvolve?
Sure! Many ways to set the interaction test, and one approach is the following with 9 rows in -gltsym:
1*Session1Condition1 - 1*Session1Condition2 -1*Session2Condition1 + 1*Session2Condition2 \
1*Session1Condition1 - 1*Session1Condition3 -1*Session2Condition1 + 1*Session2Condition3 \
1*Session1Condition1 - 1*Session1Con
by
gang
-
AFNI Message Board
> How would one code for a contrast comparing, for example, "adults rew vs. teens rew"?
How about this?
-gltCode ? 'cond : 1*rew group : 1*adults -1*teens' \
by
gang
-
AFNI Message Board
Michael,
These are the steps:
1) run 3dMean to get the mean for each condition separately, using all the subjects' data as input for that condition;
2) 3dMean -sd to obtain standard deviation for each condition separately, agin using all subjects' data that condition;
3) 3dcalc to standardize the data for each condition and for each subject separately:
3dcalc -a MeanFrom
by
gang
-
AFNI Message Board
Hi Chuck,
The script runs fine on my computer. Two comments:
1) Do you have both R packages "nlme" and "contrast" installed on your computer? If not, that could have caused the problem.
2) It looks like subject 10888 has only condition 'rew' available in the script? If the subject really has missing data, it's fine to include the subject in the script,
by
gang
-
AFNI Message Board
George, the image attachment didn't work. Do the two groups have roughly the same average age? Also, how did you center the age variable when you ran 3dttest++? Could you paste the whole script here?
by
gang
-
AFNI Message Board
Chuck, make sure you have the most recent version of AFNI. If you do, could you upload all the input files plus the 3dLME script?
by
gang
-
AFNI Message Board
You may start with plotting only the voxel with the highest PSCat group level, then gradually increase the ROI size with the radius around the peak voxel. And see what happens.
by
gang
-
AFNI Message Board
Chuck,
I'm not so sure if the following line caused the error,
-num_glt 3 \
Change it to
-num_glt 12 \
and see if 3dLME is happy.
by
gang
-
AFNI Message Board
George, it's hard to tell what went wrong without access to your data. However, you don't really need to use 3dRegAna which is outdated and *not* supported any more. Instead go with 3dttest++.
by
gang
-
AFNI Message Board
> Since I am getting PSCs from areas that have significant differences between
> groups from voxel-wise analysis, I expect to see similar kind of significant
> differences between groups but I am not. p value should be <0.05 but it is
> around 0.8. What could be wrong?
The difference in significance between the voxel-wise and ROI analysis seems to occur quite often. One possi
by
gang
-
AFNI Message Board
I don't have your mask here, but a back-of-the-envelope calculation with the following script,
3dClustSim -nxyz 39 39 40 -dxyz 3 3 3 -fwhm 6
shows the following table:
# CLUSTER SIZE THRESHOLD(pthr,alpha) in Voxels
# -NN 1 | alpha = Prob(Cluster >= given size)
# pthr | 0.100 0.050 0.020 0.010
# ------ | ------ ------ ------ ------
0.020000 55.7 62.8 71.4 78.7
by
gang
-
AFNI Message Board
> Can I assume that if the lowest q value in this dataset is 0.2608 it means that
> nothing would meet a q = .1 or q = .05 thresholding?
Sadly that's true.
Did you use a mask when you ran 3dANOVA? If not, a mask would help to some extent.
by
gang
-
AFNI Message Board
Some comments:
1) Make sure to update your AFNI with the following command:
@update.afni.binaries -d
2) AlphaSim is ancient history! 3dClustSim is much more efficient and user-friendly.
3) Forget about 3dFDR and the associated z-scores. Some AFNI programs (e.g., 3dDeconvolve, 3dANOVA*, 3dttest++, 3dMEMA, 3dMVM, 3dLME, etc.) should automatically include FDR q-value in the output, and t
by
gang
-
AFNI Message Board
> Yes, if you could add that option it would be great. Why would one consider session effect as random and not fixed?
I've modified the program so that the user can specify the fixed effects. And a variable can be modeled with either fixed or random effects, or both.
Currently the fixed effects are not provided in the output.
The modifications will be available in the next AFNI
by
gang
-
AFNI Message Board
Hi Emily,
> We ran 3dDeconvolve on each subjects' by-session data.
By this do you mean that you analyzed each session (or time point) separately or ran 3dDeconvolve with all sessions combined?
> We started with the idea of doing a 2-way, fixed effect model analysis ANOVA
> (using 3dANOVA2) with time (4 time points) as one factor and condition (4 conditions)
> as the
by
gang
-
AFNI Message Board
Hi Michael,
> 1)Why is session not considered a fixed effect?
Nobody had ever before asked for the option of fixed effects in the model. I can add such an option soon.
> Does 3dICC_REML only include the (2, 1) ICC model and not the (3, 1) ICC model?
The ICC computation is an extension of the traditional methods such as ICC(2,1) and ICC(3,1). See more discussion in Chen, G., Saa
by
gang
-
AFNI Message Board