users > detect_adni_phantom missing in 2.2.2
Showing 1-9 of 9 posts
Display:
Results per page:
Jul 17, 2012  08:07 PM | Junqian Xu
detect_adni_phantom missing in 2.2.2
Hi Torsten,

We read your "Structural MRI Unwarping Using CMTK" (realease 1.10) in the ITK Journal. This document refers to `cmtk detect_adni_phantom` in section 3.3

However detect_adni_phantom is missing in software version 2.2.2-1 (distributed by NeuroDebian). Please clarify.

Thanks
Gordon
Jul 18, 2012  09:07 AM | Torsten Rohlfing - Google LLC
RE: detect_adni_phantom missing in 2.2.2
Hi Gordon:

I am going to check with the NeuroDebian maintainer, who is the one making this particular build. One possible explanation is that CMTK may be configured without FFTW support, which would also make the phantom detection tool disappear.

Will let you know once I hear back from them.

In the meantime, you could build CMTK from source, which should be fairly straightforward. Just make sure FFTW (CMTK_USE_FFTW) support is turned on in CMake.

Best,
  Torsten
Jul 18, 2012  11:07 AM | Torsten Rohlfing - Google LLC
RE: detect_adni_phantom missing in 2.2.2
Gordon:

So indeed, CMTK was missing FFTW support on NeuroDebian. That was fixed and the missing tools should appear shortly in cmtk_2.2.2-2.

Best,
  Torsten
Jul 20, 2012  01:07 PM | Torsten Rohlfing - Google LLC
RE: detect_adni_phantom missing in 2.2.2
Okay, according to the NeuroDebian maintainers, the updated version of CMTK with the phantom detection tool is now available.

TR
Jul 22, 2012  12:07 PM | Junqian Xu
RE: detect_adni_phantom missing in 2.2.2
Thanks for the prompt response. Confirmed that detect_adni_phantom is available in 2.2.2-2  distributed through NeuroDebian.

However unwarp_phantom_image is missing (there is unwarp_image_phantom though). Might be another configure issue?

Gordon
Jul 22, 2012  02:07 PM | Torsten Rohlfing
RE: detect_adni_phantom missing in 2.2.2
Nah, this is just me using the wrong name in the IJ paper. The "unwarp_image_phantom" tool is the one you want.

Sorry - will fix that in the documentation when I get a chance.

Originally posted by Junqian Xu:

However unwarp_phantom_image is missing (there is unwarp_image_phantom though). Might be another configure issue?
Jul 22, 2012  09:07 PM | Junqian Xu
RE: detect_adni_phantom missing in 2.2.2
Also, the order of first two inputs is wrong in the IJ paper, should be

cmtk unwarp_image_phantom --final-cp-spacing 40 --levels 2 phantom.xml  phantom.nii  unwarp.xform

instead of 

cmtk unwarp_image_phantom --final-cp-spacing 40 --levels 2 phantom.nii  phantom.xml  unwarp.xform


btw, a separate question, what would cause

snr: inf
cnr: nan nan nan inf

at the beginning of the phantom.xml file

ThanksGordon
Jul 22, 2012  11:07 PM | Torsten Rohlfing
RE: detect_adni_phantom missing in 2.2.2
Originally posted by Junqian Xu:
Also, the order of first two inputs is wrong in the IJ paper, should be

cmtk unwarp_image_phantom --final-cp-spacing 40 --levels 2 phantom.xml  phantom.nii  unwarp.xform

instead of 

cmtk unwarp_image_phantom --final-cp-spacing 40 --levels 2 phantom.nii  phantom.xml  unwarp.xform


Boy, I was really paying attention when I wrote that paper, didn't I? ;)

Thanks - I'll fix that too.
btw, a separate question, what would cause

snr: inf
cnr: nan nan nan inf

at the beginning of the phantom.xml file

Best guess - the SNR and CNR spheres were not found. In that case, the entire phantom detection is not valid.

Would you mind sending me the image file that produces this result? That would be the most efficient way for me to track down what exactly is going on.

In general (and I hope this is mentioned in the IJ paper at least) the phantom detection is still a bit work in progress, so things like sanity checks may still be lacking somewhat.

Best,
  Torsten
Jul 24, 2012  12:07 PM | Torsten Rohlfing - Google LLC
RE: detect_adni_phantom missing in 2.2.2
So to wrap up this thread for the interested public -  it turns out that Inf/NaN values for SNR and CNR will arise when the dynamic range of the input images is truncated, such that all CNR and the SNR sphere have identical and constant intensity values (e.g., 4095 if 12bits are stored but the actual values pre-DICOM were significantly higher).

In that case, SNR is Inf because noise is computed as zero (all values are 4095), and the CNR NaN values arise from zero contrast divided by zero noise.

Torsten