[MINC-users] loss of image origin while converting from dicom to minc format

[Cecile Madjar]

Good afternoon,

I am running into some troubles while converting MR images from dicom to the minc format. It looks like the origin of the image is lost when converting using dcm2mnc. I tried with several version of dcm2mnc (see below the email) and this was consistent across the versions I had. 

During the conversion process, I get this warning but don't know if this could be related:
WARNING: Coordinate spacing (7) differs from DICOM slice spacing (7.98)
 (perhaps you should consider the -usecoordinates option)
 Using 7.98 for the slice spacing value.

Out of desperation, I tried considering the -usecoordinates option but I have the same origin loss issue and it uses very strange slice spacing for some acquisitions (i.e. -14.34 instead of 1.6).

I also tried converting the data from dicom to nifti and then from nifti to minc which *solves* the origin issue (in a very dirty way) but brings other issues as many fields are lost in the process (like acquisition:repetition_time and most of other acquisition:<attr> parameters). This is as problematic as losing the origin as I also need those information for further analyses.

Have anyone else experience this issue before? Is there any clean way to keep the origin of the image in the minc file? 
I will be happy to provide a set of data if necessary or any information that I may have forgotten. 

Thank you!!
Best,

Cécile 


##################################################
I tried with several versions of dcm2mnc, including:

- the latest Mac version (on Mountain Lion)
program: 2.0.07 built Nov 13 2012 18:01:11
libminc: 2.2.00
netcdf : 4.2.1 of Nov 13 2012 17:59:00 $
HDF5   : 1.8.8
 
- the version from ~April 2012 on Ubuntu (don't have the information for this binary anymore)

- the version from Aug. 2012 on Ubuntu
program: 2.0.07 built Aug  9 2012 12:44:38
libminc: 2.1.10
netcdf : 4.1.1 of Nov  7 2011 11:35:16 $
HDF5   : 1.8.4