Dear CMTK users and developers,

I tried install CMTK-3.1.0-Windows-x86.exe on my Win XP and got an error message when I run one of the command from toolkit (see screenshot). Can you help me understand the problem. Which additional libriaries I need to insall to run toolkit.

Thanks in advance.

Hi Oleg -

I don't see a screenshot actually. Did you forget the attachment?

Thanks.

Torsten

Hi Torsten,

Thanks for reply. For some reasons I change to linux.

1. I unpacked CMTK-3.1.0-Linux-x86_64 in my local folder;
2. I found .../bin and script 'correct_dwi_distortion';
3. I copied blip_up_b0 and blip_down_b0 NIFTI images in .../bin; also I add blip_down_dw;
4. I changed variable in script 'correct_dwi_distortion' and 'cmtk_functions.sh' CMTK_BINARY_DIR=/data/user/oposn/Software/DiCo/CMTK-3.1.0-Linux-x86_64/usr/local/bin
5. I run script 'correct_dwi_distortion' with arguments and get message:

/data/user/oposn/Software/DiCo/CMTK-3.1.0-Linux-x86_64/usr/local/bin$ ./correct_dwi_distortion . blip_up0000.nii blip_down0000.nii blip_down0001.nii
/data/user/oposn/Software/DiCo/CMTK-3.1.0-Linux-x86_64/usr/local/bin/cmtk_functions.sh: 38: source: not found
./correct_dwi_distortion: 69: CMTK_needs_update_and_lock: not found
./correct_dwi_distortion: 77: CMTK_needs_update_and_lock: not found
./correct_dwi_distortion: 88: CMTK_needs_update_and_lock: not found

I puzzled what to do. Please help me to overcome the problem.

Regards, Oleg

Oleg:

This looks like a packaging problem / bug in my CMake configuration.

Will take a look, but can't say when I will have time to fix this.

Thanks!

Torsten

Originally posted by Posna Oleg:

By the way Oleg -

I was actually going to retire that script (and it's sibling, correct_dwi_distortion_and_motion) from the next CMTK release.

Truth be told, the scripts don't do a terribly good job, but without much more knowledge about the input data and what happens with the output, a better script is not really doable.

As a concrete example -- we are using CMTK for most pipeline processing in NCANDA (see http://nitrc.org/projects/ncanda-datacor... We started using the CMTK default DWI scripts, but soon discovered that things like excluding bad slices (signal drop due to motion) very much depends on how you process after correction. In out case, we use Camino for tensor reconstruction, which allows us to mark any bad pixel with a negative value, to exclude that pixel for that diffusion gradient direction from the tensor fitting. That's important because after motion correction, bad slices may no longer be slices, but are likely oblique.

Anyway, long story short - if we mark bad slices with negative values, then the correction scripts need to exclude them from registration as well as from reformatting. We also noticed that, to avoid extrapolation, we need to reconstruct volumes of actual effective interpolation weights used at each pixel during motion correction.

All this makes the DWI correction scripts a lot more complex (I think the NCANDA script derived from the CMTK script is about 3x as long as the original), so and it's very specific to using Camino for recon. So it's a bit pointless to distribute such a specific script with CMTK, but when you compare the results, the difference between the quality of the tensor data using the specialized script and what is (or perhaps soon, used to be) in CMTK is like day and night.

So bottom line - I felt that we should not be distributing scripts that do a poor job by necessity of being general, and should rather not distribute scripts at all.

Will be happy to reconsider, though, if you really depend on these.

Best,
  Torsten

Thanks, Torsten!

I understood that script has a problem. Then my question is:

I have two b0 EPI 3D-images with opposite phase encoding blips. I have a series of DW EPIs (b1000). 
Which command I have to use to correct distortions? (motion is not an issue because valunteer was trained not to move)
I work in Linux environment.

Thanks again for supporting,
Regards, Oleg

Oleg:

See Section 2.3 in this document:

https://www.nitrc.org/docman/view.php/21...

Hope this helps.

Now that I look at this, since the correction scripts are documented there, I should probably not retire them after all ;)

Cheers!
  Torsten

Dear Torsten,

I followed Section 2.3 and got a message:

bash: cmtk: command not found


In the /bin there is no cmtk executable. When I deleted cmtk and started
./epiunwarp
then I got
./epiunwarp: error while loading shared libraries: liblzmadec.so.0: cannot open shared object file: Nosuch file or directory


Sorry for bothering you again. Thanks a lot for the reply.

Regards, Oleg

I am going to call this a bug:

http://www.nitrc.org/tracker/?func=detai...

At this point, though, I cannot offer an estimate for when it might get fixed.

TR

This one's a little simpler -

Try this preview of the upcoming 3.2.0 release:

https://www.nitrc.org/frs/download.php/6...

I disabled some optional features to reduce the number of external libraries required (especially "esoteric" ones).

You may still need to have the "fftw3" library installed - disabling it would remove important functions from CMTK, and it's an easily avalilable package on pretty much all Linux distros, so it shouldn't cause any trouble.

Best,
  Torsten