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 3991 - 4020 of 4524
I think NIfTI datasets should work fine.
Did you try it and have some problem?
- rick
by
rick reynolds
-
AFNI Message Board
Hi Basile,
Obliqueness is not handled by 3dresample, which works as
the afni GUI does.
If you want to make the obliqueness of one dataset match
that of another, consider "3dWarp -oblique_parent".
- rick
by
rick reynolds
-
AFNI Message Board
When interpolating masks, we either use NN or voting
(as in 3dfractionize), so that the mask values are preserved.
It's pretty wimpy when compared with splines.
- rick
by
rick reynolds
-
AFNI Message Board
You are computing the area under the curve of the
estimated BOLD response. Do not threshold on such
a volume. Of course the values outside the brain will
dwarf those inside.
Typical values will probably be between 0.1 and 1,
depending on the duration of your stimulus (for an
"active" voxel). Values outside the brain will often
be much higher, but not significant.
- rick
by
rick reynolds
-
AFNI Message Board
The stim_type column was added to uber_subject.py at the
beginning of October.
If your stim durations are all above 15 seconds say, then
convolutions of the durations will basically not change
its amplitude. That is good, and makes your life easier.
In such as case, just use -stim_times_AM1 (if your only
modulation is duration) and probably the dmBLOCK(1) basis
function.
To get an
by
rick reynolds
-
AFNI Message Board
Hi Basile,
Blocky is supposed to be somewhere between NN and linear
interpolation. Like linear, it is restricted to the first
neighbors. But while linear gives weights that are simply
the fractional distances to neighbors, blocky gives weights
that are akin to x^4, in that being closer to one neighbor
gives that neighbor much more (of the fractional) weight.
I don't really know
by
rick reynolds
-
AFNI Message Board
Hi Danny,
The Coef volumes are the beta weights, so those are likely
what you will run group analysis on.
Percent signal changes will always be very large outside
the brain of a scaled dataset. But as you see, they are
not very significant (and will eventually be masked by the
group mask). You generally do not threshold on the betas.
To view them, consider opening a separate contro
by
rick reynolds
-
AFNI Message Board
It might be a good idea to describe the range of
durations you are talking about. Are they between
3 and 5 seconds, or between 2 and 20, say?
Note that the amplitude modulators are specified
in the timing file. For details, please see the
"Amplitude Modulated FMRI regression analysis"
section at
2007 3dDeconvolve updates
- rick
by
rick reynolds
-
AFNI Message Board
Hi Lauren,
We have not yet decided on the schedule of classes, so Friday
is still up in the air.
We are also considering not even teaching the Unix class, since
there will be so many people. I may write up a tutorial to get
people started, instead.
Sorry not to be more definitive, but that is where we are right now.
- rick
by
rick reynolds
-
AFNI Message Board
Hi Rengin,
Also, what are the grid sizes for those datasets? I assume that the
tlrc grid is higher resolution than the orig grid, leading to a larger voxel
count. Is that the case?
Though even so, that should make the required cluster sizes of smaller
volume in tlrc space (more voxels, but smaller volume), making it
(unfairly) easier to get results in the standard space case.
In ge
by
rick reynolds
-
AFNI Message Board
Hi Sarah,
There isn't much to it, but we do not yet have a 10.8 machine
to be sure. The expected steps are (assuming you use tcsh):
1. install the macosx_10.7 binaries under $HOME/abin
2. in .cshrc, add $HOME/abin to your path
set path = ( $path $HOME/abin )
3. also in .cshrc, set DYLD_FALLBACK_LIBRARY_PATH to your abin
setenv DYLD_FALLBACK_LIBRARY_PATH $HOME/abin
We w
by
rick reynolds
-
AFNI Message Board
Hi Liesel,
Since you say R_io.so is in your abin, yet 3dMVM is not
finding it, make sure that DYLD_FALLBACK_LIBRARY_PATH is
set to your abin directory. For example, in your .cshrc file, put:
setenv DYLD_FALLBACK_LIBRARY_PATH $HOME/abin
Assuming you are using tcsh (and not bash), you could try
that at the command line and then rerun the 3dMVM script.
- rick
by
rick reynolds
-
AFNI Message Board
Hi Chris,
The examples with indices are the more advanced ones, for when
you are using multiple basis functions per stim class (e.g.
using 8 TENTs, and summing lags 2..5 (skipping 0,1,6,7), as in
the example with +Ear[2..5]).
Your case is simple, do not use any indices.
-gltsym 'SYM: Ear -Finger' -glt_label 1 EvsF
Note that there is no space between '-' and
by
rick reynolds
-
AFNI Message Board
Hi Venessa,
The first command seems to be running 3dmaskave on a beta
data, meaning the result would be a single value, not a
time series.
That command should be run on the EPI time series that
would otherwise be input to 3dDeconvolve, say.
- rick
by
rick reynolds
-
AFNI Message Board
Hi Gaurav,
One would certainly want to cluster at the group level.
High significance of individual subject results is actually
irrelevant (usually) in the group analysis. All it means is
that a given subject beta that goes into the group analysis
is reliable.
What approach 1 would do is exclude subject voxels from the
group test that are not sufficiently reliable, limiting the
numb
by
rick reynolds
-
AFNI Message Board
I'm not sure if one can do that in the GUI, however
it is easy by driving afni from the command line
(when starting afni with -yesplugouts or turning them
on in the GUI after startup). Consider this:
plugout_drive -com "SET_THRESHNEW A 0.001 *p" -quit
- rick
by
rick reynolds
-
AFNI Message Board
Then the DOF are probably varying across subjects,
which is to say, you are probably censoring (as I should
have expected to begin with).
Running '3dinfo -verb' on the stats dataset should show
you the applied DOF. For example, on a t-stat volume,
you might see: statcode = fitt; statpar = 183, indicating
183 DOF.
- rick
by
rick reynolds
-
AFNI Message Board
Hi Gaurav,
You wouldn't exactly do that on the stats dataset. The only thing you
need to know (which is indeed contained in the stats dataset) is the
DOF (degrees of freedom) value to apply. E.g. if dof = 184 (from 200
TRs and 16 regressors), then try:
cdf -p2t fitt 0.001 184
- rick
by
rick reynolds
-
AFNI Message Board
This may end up being offered up to twice per year,
including later in 2013.
- rick
by
rick reynolds
-
AFNI Message Board
Hi Matt,
The TRs per stim are the TRs where the expected BOLD response
is non-zero. Particularly in the case of an event-related design with
only a few classes, most of the stim classes could have non-zero
ideals for most of the TRs.
I need to add some help pages (either in afni_proc.py or in
gen_ss_review_scripts.py) for that stuff to accurately describe each
of the outputs...
-
by
rick reynolds
-
AFNI Message Board
Hi Peter,
Thanks for the details.
What is happening is not quite obvious to me, but I guess that
you are at least getting the extents mask applied in the volreg
block. Consider adding the scaling step, just to see whether
that has an effect. Scaling should not affect the statistics,
aside from the beta magnitudes. And it is just a test.
The reason I keep asking about the extents m
by
rick reynolds
-
AFNI Message Board
Hi Kausar,
Those exact options are working fine for me. What
version of AFNI do you have? (2 lines from afni -ver)
This was added to afni_proc.py in September.
- rick
by
rick reynolds
-
AFNI Message Board
Hi Peter,
0. What statistic is showing this (e.g. Full-F or one for a specific class)?
1. Did you run the 3dREMLfit command via -regress_reml_exec in AP.py?
2. Was the extents mask applied in the scaling block?
3. Is there anything special with how you defined what goes into
the baseline part of the model (via -stim_base options)?
4. Which motion terms were used in the model?
It is
by
rick reynolds
-
AFNI Message Board
Re: FDR - 11 years ago
Hi Linda,
In the case of -pmask and -list, the p-values of all voxels are
printed out, not just those in the mask. However, all of the
voxels not in the mask should have a q-value of 1.000000 and can
be ignored.
The -pmask option does not duplicate the more useful functionality
of -mask. So it might be better to use the -mask option. You
would then get the same list as with -pmask,
by
rick reynolds
-
AFNI Message Board
Hi Chris,
I don't know, that is something that you should
preferably ask your physicist(s) or scan tech(s).
If you believe that it is alt+z, you should at least
be able to see the 'alt' part of it just in the viewer.
Open a couple of windows in afni and view the
first pre-steady state TR as the underlay.
Doing that, do you see a slice-wise striping pattern
(i.e. alter
by
rick reynolds
-
AFNI Message Board
Hi JoJo,
Perhaps the most likely example would be to use the XYZAVE method,
which is to say, use "-batch XYZAVE commandfile".
The format of commandfile depends on the method. For the XYZAVE
method, the commandfile format is a sequence of seed lines, where
each seed line is of the form "dset_prefix x y z". So the file
might look like:
M_PCC_dset 0 54 26
by
rick reynolds
-
AFNI Message Board
Hi Rengin,
Having the eyes show up that way isn't quite what I was
expecting. This difference is in some ways not very
concerning, and in others, it is. What you show is not
a big difference between blurs or registration, say. It
is a small but clearly notable difference between the
regression models.
Focus on the regressors, not on the preprocessing steps.
And control over the
by
rick reynolds
-
AFNI Message Board
Hi Rengin,
It is not strange to have results outside the brain, but
it would be good to understand any difference. The
only the question about registrations is whether it
worked well.
Without being able to look at your data and any
difference in the results, it is very hard to say much.
When you look at both results at the same time, are
they shifted versions of each other (approx
by
rick reynolds
-
AFNI Message Board
What are you looking at to make the determination,
voxels that are significant? Because the main thing
that I don't see (besides registration) is a scale step
in your script.
However, you are also using -1filter_blur 6, instead
of -1blur_fwhm 6. I wonder if your script is not actually
applying any blur. That would certainly account for
more widespread activation from the afni_pro
by
rick reynolds
-
AFNI Message Board
Hi Ana,
Since you are using the straight 10.6 binaries, you should
be getting glib via fink.
What is the output of "fink list glib"?
- rick
by
rick reynolds
-
AFNI Message Board