open-discussion > DICOM to NRRD
Showing 1-17 of 17 posts
Display:
Results per page:
Jul 21, 2017  08:07 AM | James Hartzell
DICOM to NRRD
Hello

Do you have updated instructions for converting DICOMS to NRRD?  
I've downloaded the latest versions of DTIPrep and Slicer, but the various earlier versions of instructions on this list and
on the Slicer list don't seem to work for me.
Jul 21, 2017  11:07 AM | Martin Styner
RE: DICOM to NRRD
Best use the DWIConverter module in Slicer, either through its commandline interface (DWIConvert)  or through Slicer. Here is a description as well as nice tutorial:
https://www.slicer.org/wiki/Documentatio...

Martin
Jul 24, 2017  01:07 AM | James Hartzell
RE: DICOM to NRRD
Thanks very much Martin--the instructions work quite well, though they left out the 'save' step at the end, and the current Slicer menu structure is a little bit different under the conversion tab.
Cheers
Jul 27, 2017  06:07 AM | James Hartzell
RE: DICOM to NRRD
HI Martin

I have a follow-up question.  When I load the default protocol for the NRRD DWI's
I get the following message:
The following programs have not been found.
Please enter the path manually if necessary:
> bet2
> DiffusionWeightedVolumeMasking

Is it necessary to have the nodif_mask for running DTIPrep successfully? 
I ask because I already have made masks using bet2, and for some reason I can't find the path to Slicer's DiffusionWeightedVolumeMasking

Cheers
James
Jul 27, 2017  07:07 AM | Martin Styner
RE: DICOM to NRRD
Hi James
No, you do not need those, unless you are using DTIPrep to run bet2 (in your case, you have done that already, so no need to redo it) or you want to use the Slicer based DiffusionWeightedVolumeMasking (again, not needed in your case). Thus you can safely disregard these warnings.

Best
Martin
Jul 27, 2017  07:07 AM | James Hartzell
RE: DICOM to NRRD
Hi Martin

Thanks very much---just wanted to be certain on this point.

Cheers
James
Aug 2, 2017  05:08 AM | James Hartzell
RE: DICOM to NRRD
HI Martin

Could you point me to clear instructions for converting the resulting QCed.nrrd files back to nifti (nii.gz)?

Cheers
James
Aug 3, 2017  01:08 AM | James Hartzell
RE: DICOM to NRRD
Hi Martin

A colleague guided me on the conversion from NRRD to NIFTI using Slicer.

I notice though that the QCed bvecs have flipped the x and z axis values, so that negative became positive and
positive became negative.  Is there some way to prevent this from happening, since otherwise I have to manually
correct the bvalue texts.

Cheers
James
Aug 20, 2017  09:08 AM | Martin Styner
RE: DICOM to NRRD
Hi James

Are the bvecs fully inverted (i.e. multiplied by (-1)), or is it only the x and z components? If all components are inverted, then you do not need edit the bvecs (as diffusion MRI is not sensitive to inversions of the gradient direction).

Otherwise, this would be a bug needing to be reported to the conversion tool (DWIConvert), best on its github page.

Martin
Aug 20, 2017  10:08 AM | James Hartzell
RE: DICOM to NRRD
HI Martin

Thanks for the feedback.
The sign inversion is just in the x and z coordinates. I posted to the Slicer list and they responded that it is a bug in DWIConvert that they're implementing a fix for.
Cheers
James
Jan 11, 2018  07:01 AM | Daniel Berge - Hospital del Mar Medical Research Institute (IMIM)
RE: DICOM to NRRD
Hi James and Martin  I had a similar problem than James converting nrrd files to FSL. Using the latest version of DWIConvert and flag ---outputNiftiFile instead of--outputVolume in Slicer 4.8, it wirtes bvec file with gradients directions exactly as in nifti. However this works with standard nrrd files, but not for preprocessed DTIPrep nrrd files (*_QCed.nrrd). It reports that the only supported output is "short" . There is an "unsigned short" in the headers of *QCed.nrrd instead of "short" in the regular nrrd files.
 Any suggestions of how to solve this would be very appreciated.

I am attaching an example of this QCed.nrrd file in case it helps

Thanks
Jan 11, 2018  07:01 AM | Martin Styner
RE: DICOM to NRRD
Not sure what the issue us with the updated DWIConvert (i.e. why it would not take unsigned short data). Assuming that your DWI data does not have intensity values above 32,000 , you could simply convert the QCed unsigned short DWI file to a signed short one using unu (a swiss-army-knife style utility for NRRD that is packaged with Slicer):
unu convert -t short -clamp -i inputUnsignedShort.nrrd -o outputShort.nrrd

Best
Martin
Jan 11, 2018  07:01 AM | Daniel Berge - Hospital del Mar Medical Research Institute (IMIM)
RE: DICOM to NRRD
Thanks! I also noticed that qced file header says format gzip instead of raw as in regular. Nrrd. Any chance that this has something to do with this issue?
Thanks again.
Jan 11, 2018  07:01 AM | Martin Styner
RE: DICOM to NRRD
no, zipping the data in nrrd is standard, so that should not be an issue. 

One note though: unu will write raw data, if you want unu to write zipped data you need to also use it's save option, e.g.

unu convert -t short -i input.nrrd | unu save -e gzip -f nrrd -o output.nrrd

Martin
Jan 11, 2018  12:01 PM | Daniel Berge - Hospital del Mar Medical Research Institute (IMIM)
RE: DICOM to NRRD
> Thanks a lot! That worked perfectly! Just in case any other uses this,
> unu can be found in Slicer 4.4 and previous versions, launched as
> ./Slicer --launch unu convert( and flags and options mentioned by
> Martin Styner in thread)

Mar 22, 2018  02:03 PM | Micah Johnson - UCLA Department of Psychology
RE: DICOM to NRRD
Hi James, do you know if this has been fixed? (or could you point me to any further discussion on it?)

I'm having the opposite problem (using Slicer 4.8)...only my y axis is flipped.

Many thanks,
Micah

Originally posted by James Hartzell:
HI Martin

Thanks for the feedback.
The sign inversion is just in the x and z coordinates. I posted to the Slicer list and they responded that it is a bug in DWIConvert that they're implementing a fix for.
Cheers
James
Mar 27, 2018  05:03 AM | Maike Kleemeyer
RE: DICOM to NRRD
Dear all,

do you have any ideas on how to solve this issue when running on Mac OS? Is there an option for outpuuting DTIPrep data NOT in unsigned short type?

Thanks, Maike