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
otto Wrote:
-------------------------------------------------------
> It is a block analysis
> where I expose subjects to the following paradigm:
> warm, cold, warm, cold temperature (240s each).
>
> […]
> I also measured the skin temperature of the
> subjects which has the basic
> shape of a sine curve (warm is up and cold is
> down).
In this design you
by
nick
-
AFNI Message Board
otto Wrote:
-------------------------------------------------------
> What I really want to do is to correct the time
> series only for motion and scanner drift,
> e.g. regress the motion parameters and the
> polynomial baseline without any "regressor
> of interest" (first 12 columns in X matrix).
How about using 3dDetrend?
by
nick
-
AFNI Message Board
Noyan T Wrote:
-------------------------------------------------------
> The beta estimate values I'm looking for would be
> the #0_Coef values in AFNI, right?
Yes indeed.
by
nick
-
AFNI Message Board
otto Wrote:
-------------------------------------------------------
> What does errts
> stand for?
"error time series". It is another way of saying residuals from the GLM, as in the GLM
y = X * b + e (where y=data, X=design matrix, b=beta estimates, e=residuals),
the residuals can be expressed as e = y - X*b. As such they indicate the 'error' of the model (
by
nick
-
AFNI Message Board
Noyan T Wrote:
-------------------------------------------------------
> How do I obtain the beta estimates for each
> condition?
Typically with N times using '-stim_times ...', one for each of the N conditions. Also don't forget to set
'-num_stimts'. Further also a good idea to set -stim-label.
> Do I make a -gltsym for each condition?
That is not
by
nick
-
AFNI Message Board
Noyan T Wrote:
-------------------------------------------------------
> Ideally, the end goal is to create a generative
> model that ties the stimuli to the response so
> that a novel stimuli can be used to predict a
> response and vice versa.
>
> Given that end goal, I think using the GLM for
> noise reduction and then using other tools to
> learn the mapping i
by
nick
-
AFNI Message Board
Noyan T Wrote:
-------------------------------------------------------
> I want to contrast each condition to each other.
With 60 conditions that makes 1,770 comparisons, which is a lot. How are you planning to interpret these?
You may consider to reduce the number of comparisons. Is there any structure in the experimental design?
> Also, I think the TENT function is the most
>
by
nick
-
AFNI Message Board
Noyan T Wrote:
-------------------------------------------------------
> I've only used 3dDeconvolve for block designs
> before, so there's the BLOCK(x,y) parameter. What
> do I replace that with?
In the 3dDeconvolve help, you can read that " you can use 'BLOCK' for event-related analyses".
Also I suggest you to read the help further to make an info
by
nick
-
AFNI Message Board
Noyan T Wrote:
-------------------------------------------------------
> I'm trying to understand the 3dDeconvolve program
> but I'm not sure how it applies to my case.
>
> I have N types of stimuli presented at
> pseudorandom time intervals, each stimulus
> presented k times. I'm trying to fit a GLM
> to my bold response to eliminate noise.
Only t
by
nick
-
AFNI Message Board
Hi Bob,
Bob Cox Wrote:
-------------------------------------------------------
> I suppose that wouldn't be too hard.
>
> .....
>
> 7 minutes later -- source code is changed. Will
> recompile binaries tonight.
thanks, that's much appreciated!
best,
Nick
by
nick
-
AFNI Message Board
dennisl Wrote:
-------------------------------------------------------
> I am trying to run 3dProbTrackID with some HARDI
> scans after successfully doing so with DTI but I
> get a FATAL ERROR:[…]
>
> I am not sure what "mask volumes" is referring to?
Those seem to be the volumes provided with "-uncert" and "-netrois".
Can you run 3dinfo o
by
nick
-
AFNI Message Board
Hello Bob et al,
We have our own AFNI-based pipeline in the lab and found that sometimes 3dDespike with the standard auto mask removes too many voxels for certain subsequent (between-participant) analyses.
Looking at the source code [1] it seems that around line 380, 3dDespike first applies an auto mask and then dilates it 4 times.
Would you be willing to make this currently hard-coded
by
nick
-
AFNI Message Board
vgb88 Wrote:
-------------------------------------------------------
> I use 3dSurfMask as a surrogate for 3dautomask and
> the program works great except for one minor
> glitch. Once the mask is generated and saved, the
> program does not terminate for some reason. This
> makes scripting very difficult!
Could you provide the *exact* command you are using with 3dSurfMask?
by
nick
-
AFNI Message Board
Hi Emily,
Emily Wrote:
-------------------------------------------------------
> Here is the full output: […]
> *+ WARNING: Disk space: writing file
> ./10017.GM_XYZ_OYmeqfZ3JqF7n9LNC7gDAA+orig.BRIK
> (16 MB), but only 0 free MB on disk
It seems your hard drive is full…
You can check this with "df -h".
best,
Nick
by
nick
-
AFNI Message Board
v.iacovella Wrote:
-------------------------------------------------------
> in topic, I was trying to understand all the steps
> within
> this afni script to align
> two EPI sessions (for example in a longitudinal
> fmri study).
>
> I do not get why one should extract the square
> root of the first affine matrix. Is there a
> mathematical reason?
The rati
by
nick
-
AFNI Message Board
PeterKohler Wrote:
-------------------------------------------------------
> Are .gii, .1D
> and .niml.dset files essentially equivalent, as
> far as SUMA is concerned?
Kinda… for your purposes, probably yes.
1D is the simplest and most limited format. It is used to store just numbers, not meta-information such as labels, type of statistic, or degrees of freedom associated with
by
nick
-
AFNI Message Board
PeterK Wrote:
-------------------------------------------------------
> If I put all my runs into afni proc py like this:
>
> -dsets run1+orig run2+orig run3+orig run4+orig
> etc...
>
> with timing files where each line corresponds to
> each consecutive run (like usual), is there a way
> to get it to spit out stats (coeffs) for each run
> seperately?
You
by
nick
-
AFNI Message Board
Emily Wrote:
-------------------------------------------------------
> I have tried to incorporate -init_xform
> AUTO_CENTER into my @auto_tlrc script because my
> centers were off by 40 mm when I ran the script
> without it. I am hoping it will help better align
> my GM masks onto an MNI template.
>
> This is the error message I get: Error: Anatomical
> dset
&
by
nick
-
AFNI Message Board
Hi Emily,
Emily Wrote:
-------------------------------------------------------
> So, to remove just the baseline in each run for
> afni would I just add "regress" as a block in my
> afni proc py script? Would I need to specify
> anything? This is all I want to do, so still no
> bandpass, no motion regression, no censoring,
> etc., just remove the baseline.
by
nick
-
AFNI Message Board
Hi Daniel,
Daniel Glen Wrote:
-------------------------------------------------------
> for the original way
> with both the FreeSurfer output and the original
> datasets in orig space, @SUMA_Make_Spec_FS handles
> the shift between the Freesurfer and the original
> MRI data with @Align_Centers. The datasets should
> be identical other than a shift,
That's not
by
nick
-
AFNI Message Board
Hi Ziad,
ziad Wrote:
-------------------------------------------------------
> I'd hate to make 3dAllineate attempt alignments
> that are potentially quite far off, especially
> when the transform to MNI space should be
> available somewhere in that pipeline. If you just
> can't get your hands on it, you can go with the
> parameters you set if they work well
by
nick
-
AFNI Message Board
Dear AFNI experts,
In the passed I used @AddEdge with +orig datasets, but we've now started to move to +tlrc.
It seems that @AddEdge does not support +tlrc (see below); and I found that the script has '+orig' postfixes all over the place. Is that correct? If so, are there any plans to add +tlrc support?
Of course I can copy/refit to +orig and then apply +tlrc later, but th
by
nick
-
AFNI Message Board
Hi Emily,
Emily Wrote:
-------------------------------------------------------
> Actually, the ICA program that I use
You mentioned GIFT [1], so I assume that's what you use.
> actually will
> do the normalization and detrending steps.
I RTFM [1] (well, glanced at it) and there is mention of support for multiple sessions.
That might come in handy. The thing is,
by
nick
-
AFNI Message Board
Emily Wrote:
-------------------------------------------------------
> I am preparing some resting state data for ICA
> analysis. Per Vince Calhoun, I did not run the
> regression prior to ICA. Because I do not run the
> regression, I do not get an errts file. So I need
> to use the functional files from the last step
> done which was blurring, so I assume I would use
&g
by
nick
-
AFNI Message Board
Emily Wrote:
-------------------------------------------------------
> Just to clarify, you think that concatenating the
> two pb04.${subject}.blur+tlrc files from the afni
> proc py output using 3dtcat would be okay?
I was talking about errts files, not blur files. I assume the latter are the result from smoothing the data.
Why do you want to concatenate these files?
by
nick
-
AFNI Message Board
PeterKohler Wrote:
-------------------------------------------------------
> Surely I must be missing something, but I cannot
> figure out how to convert a .mgh file containing
> FreeSurfer surface overlay data into a niml.dset
> file that SUMA can read.
According to [1] mgh is a volumetric format, not for surfaces.
Although I've never done this, the following may wor
by
nick
-
AFNI Message Board
elisa Wrote:
-------------------------------------------------------
> in running the 3ddeconvolve script, I get a
> message that recommends a minimum polort = 4.
> However, I obtain more reasonable results
> (replicating previous literature) if I stay with
> polort = 1. What strategy do you recommend?
If polort 1 versus 4 gives you considerable different results that wou
by
nick
-
AFNI Message Board
Dear AFNI experts,
Until recently we (our lab) did our volumetric analyses in original space, but now we are switching to a standard (non-afni_proc.py) pipeline where brains are aligned to an MNI template quite early.
In our surface pipeline (that is quasi-independent from the volumetric one), surfaces are reconstructed using FreeSurfer based on anatomical volumes in the original space. The
by
nick
-
AFNI Message Board
Emily Wrote:
-------------------------------------------------------
> However, I realized that if I don't run these
> steps, I get two seperate runs instead of a single
> errts file. Is there anyway to concatenate these
> files without running these steps? I have been
> using the pb04.${subject}.blur+tlrc files.
Joining data directly from different runs can be a tric
by
nick
-
AFNI Message Board
dbliss Wrote:
-------------------------------------------------------
> Am I to infer that AFNI can compare datasets
> perfectly sensibly even when their in different
> orientations? Or should I re-run everything?
> Will having datasets in different orientations
> screw things up for me farther down the road?
You should be fine. There are 3! * 2^3 = 48 possible orientat
by
nick
-
AFNI Message Board