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
I would recommend using 3dTproject for bandpassing -- it can do this in combination with censoring and other nuisance regression.
by
Bob Cox
-
AFNI Message Board
What kind of "strange results"? It is important to skull strip the brain image if aligning to a skull stripped template -- as an example of something that could give unpleasant results. Or to NOT skull strip the image if aligning to a whole head template.
If something is already in TT space, then it is close to MNI space already. In that case, a command like
3dQwarp -base ~/abi
by
Bob Cox
-
AFNI Message Board
Actually, the time derivative of a function is nearly orthogonal to the function. You might have a problem with some of the regressors being nearly zero. Have you graphed them, say with
1dplot -sepscl filename.1D
You can orthonormalize a set of 1D vectors with the 1dsvd command, as in
1dsvd -1Dleft filename.1D > orthog.1D
At the top of each column is the associated singular value
by
Bob Cox
-
AFNI Message Board
To add to Rick's reply,
In any AFNI program (but not in FSL or SPM), you can operate on a subset of input volumes in a dataset using the subscript notation that Rick's reply showed -- the 'quotes' are needed since the [ and ] characters means something special to the shell you into which you are typing.
In the subscript notation, '$' means 'last volume in
by
Bob Cox
-
AFNI Message Board
A few days ago, I introduced a bug into 3dNwarpApply when "improving" the error messages. The result was that it wouldn't work at all, and would fail with a message like
Can't copy nwarp dataset or Can't extend nwarp dataset
I have now re-improved the error message checking, so that 3dNwarpApply doesn't crash every time. The binaries with this fix are now ava
by
Bob Cox
-
AFNI Message Board
This workshop in Athens (Greece, not Georgia) sounds interesting. "Connectomics" is the new word that all the cool kids are buzzing about.
https://biomedic.doc.ic.ac.uk/miccai16-bacon/
by
Bob Cox
-
AFNI Message Board
Clothed with transcendent brightness, afni now has the option
-XXXnpane P
which will set the number of color 'panes' used in the continuous color scales to 'P', which must be an even integer between 256 and 2048 (inclusive). Multiples of 256 will probably work best with the existing color maps in AFNI, which are designed for that number of panes.
This change is now
by
Bob Cox
-
AFNI Message Board
Despair not, dear ZXu. The change has been made, and it will be compiled on Monday (09 May 2016). The AFNI command line will have an option to set the number of colors to use in the "continuous" colorscale. Until then, keep up your spirits, and ponder these words of the immortal Milton:
Some glimpse of joy, to have found thir chief
Not in despair, to have found themselves not los
by
Bob Cox
-
AFNI Message Board
Dear Dr Xu (if that is your name):
What will you give us if we increment NPANE_BIG? Or make it user-settable (e.g., command line option)? Are you talking AFNI here, or SUMA?
by
Bob Cox
-
AFNI Message Board
This is a difficult questions, since the underlying "signal" model for resting state analyses is not clearly specified -- basically, we are asking if THIS part of the brain fluctuates in time sort of like THAT part of the brain. So what's the null hypothesis and what's the alternate hypothesis?
by
Bob Cox
-
AFNI Message Board
It would help if you gave the names of some of the "invisible files".
AFNI-readable datasets are stored in files whose names end in .HEAD or .nii or .nii.gz -- you won't see the other files in the directory in the AFNI GUI.
Files with names ending in .1D are text files with columns of numbers -- these are visible in the AFNI GUI, but not as 3D datasets to visualize.
by
Bob Cox
-
AFNI Message Board
All roads lead to AFNI, it would seem. In other words, AFNI is the New Rome!
by
Bob Cox
-
AFNI Message Board
At a guess, the pre-processing of the data (before 3dSetupGroupInCorr) removed so many degrees of freedom that there isn't much variation between the different voxel time series.
by
Bob Cox
-
AFNI Message Board
OMG, AFNI runs under Windows "out of the box" !?
These are the end times of the world.
by
Bob Cox
-
AFNI Message Board
The actual calculation is with the arctanh(correlation) -- quoting from the help
++ Then carry out the t-test between/among these 3D correlation maps,
possibly allowing for dataset-level covariates.
-- Actually, between the arctanh() of these maps:
cf. RA Fisher:
-- To be overly precise, if the correlation is larger than 0.999329,
t
by
Bob Cox
-
AFNI Message Board
Yes, that is what I recommend at this moment. You will find that the minimum cluster size goes up significantly over the 'classic' method.
The reason this new approach is not in afni_proc.py is that I don't want to "write into stone" (OK, Python) something that is likely to change soon.
by
Bob Cox
-
AFNI Message Board
This is a difficult question to answer, since I am working on this subject RIGHT NOW, and it remains in a state of flux. At present, the simplest change to make is to use 3dFHWMx with the -acf option. Then average those 3 auto-correlation function parameters (not the 'classic' FWHM estimates) across subjects, and use 3dClustSim also with -acf and those averaged parameters.
This gui
by
Bob Cox
-
AFNI Message Board
There is also the "InstaCalc" function, for the user at the AFNI Certified Expert level (or above).
by
Bob Cox
-
AFNI Message Board
There is the "Automask" button in an image viewer window, which you can try -- to access it, you have to popup the right-click hidden menu attached to the intensity scale bar to the right of the image, and find that button. A lot of other little-used image viewing features are relegated to this dusty attic of AFNI.
by
Bob Cox
-
AFNI Message Board
At present this cannot be done. Perhaps it could be added, but right now my time is crammed full of upcoming crises that leave me in a state of thrashing.
by
Bob Cox
-
AFNI Message Board
What does 3dDeconvolve do exactly to the censored time points - does it replace it with zeros?
The censored time points are deleted -- that is, if the original data time series has 150 points, and 3 are censored, the data time series to be analyzed will have 147 time points -- the censored points are not replaced with zeros, but actually cut out of the data. To match this, the regression matr
by
Bob Cox
-
AFNI Message Board
Here are some results from what I found to be a reasonable combination of options for registering with the "large ventricles" problem:
3dQwarp -base MNI152_T1_2009c_uni+tlrc \
-source BBB_uni+tlrc -prefix BBB_new+tlrc \
-allinfast -minpatch 15 -Qfinal -verb -pblur |& tee BBB_new.out
Link to Images
Images on top are the "old" data
by
Bob Cox
-
AFNI Message Board
3dBandpass -despike uses a simple and fast 9-point despiking function. 3dDespike (OLD or NEW) uses a more complex function that should deal better with data drifting and functional activation.
We no longer use 3dBandpass in our recommended processing stream, but rather use 3dTproject as part of the afni_proc.py script. I would recommend using that script as the basis for your dataset process
by
Bob Cox
-
AFNI Message Board
I doubt that it will make much difference which way you blur. With that assumption, I'd recommend 3dBlurInMask, since it is implemented more efficiently than the filtering options in 3dmerge.
by
Bob Cox
-
AFNI Message Board
I forgot to add the '-pblur' option. And you could try the '-noweight' option -- I don't know if it would help, but it might.
by
Bob Cox
-
AFNI Message Board
That depends on how you installed it. If you use our usual recommendations, then the binaries and plugins, etc., are stored in a directory named abin in your home directory. You can just delete that directory and its contents. If you also installed AFNI sample data, the AFNI bootcamp course material, or other material, you'll have to find where you put that and delete those directories as
by
Bob Cox
-
AFNI Message Board
It works by a 6-point finite difference numerical (approximate) solution to a diffusion equation
du/dt = D { d2u/dx2 + d2u/dy2 + d2u/dz2 }
and applying Neumann boundary conditions at the edge voxels. Which means that at an edge voxel, if a grid point needed in the finite difference scheme is outside the mask, then the reflected grid point inside the mask will be used -- and if the reflecte
by
Bob Cox
-
AFNI Message Board
"Possible" in the sense of re-writing the AFNI colorization scheme. But I'm not doing that anytime soon, considering all the other things that need doing in AFNI-land.
by
Bob Cox
-
AFNI Message Board