[Mrtrix-discussion] mrconvert DICOM file order

[Donald Tournier]

Hi Felix,

OK, it looks like the AcquisitionNumber fields are the problem. MRtrix will
use this number to sort DICOM images, ahead of anything else. The actual
sorting is done according to the code at lib/file/dicom/image.cpp:190, and
basically does:
- sort by acquisition number; if identical:
- sort by distance along normal to slice; if identical:
- sort by sequence number; if identical:
- sort by instance number

For what you sent, it looks like instance number would be more appropriate
in this case. What I don't understand is why the acquisition numbers appear
so random. Maybe I've hit the limits of my knowledge of DICOM... If anyone
can shed some light on the correct way to handle this situation, I'd love
to hear from them.

In the meantime, you could simply delete line 190
of lib/file/dicom/image.cpp, which presumably will then sort these data
correctly. The other way that wouldn't require modifying the code would be
to place each DICOM mosaic file in its own folder so that you can get
MRtrix to scan that image on its own, mrconvert each volume to a suitable
name like tmp-0001.nii, tmp-0002.nii, etc, and then do a final:
$ mrconvert tmp-[].nii combined.nii
Using the square brackets as above allows you to treat a numbered series of
3D volumes as a 4D image (see the
documentation<http://www.brain.org.au/software/mrtrix/general/cmdline.html#sequences>for
details). This approach is a bit of a kludge, but could be scripted
fairly easily.

I was pretty sure my way of handling image sorting was robust (this is the
first reported issue of this kind since release in 2008!), but obviously
I'll need to take a closer look into this when I have a minute. If anyone
has a good source of information on this, I'm all ears!

Cheers,

Donald.


On 25 July 2012 00:13, Félix C. Morency <felix.morency at gmail.com> wrote:

> Donald,
>
> Thank you for the fast answer. Here are partial informations of
> 000[1-4].dcm using read_dicom:
>
> 0001.dcm
> [DCM] 20 10 SH 2 2588 StudyID : [ 1 ]
> [DCM] 20 11 IS 2 2598 SeriesNumber : [ 4 ]
> [DCM] 20 12 IS 2 2608 AcquisitionNumber : [ 1 ]
> [DCM] 20 13 IS 2 2618 InstanceNumber : [ 1 ]
>
> 0002.dcm
> [DCM] 20 10 SH 2 2694 StudyID : [ 1 ]
> [DCM] 20 11 IS 2 2704 SeriesNumber : [ 4 ]
> [DCM] 20 12 IS 4 2714 AcquisitionNumber : [ 256 ]
> [DCM] 20 13 IS 2 2726 InstanceNumber : [ 2 ]
>
> 0003.dcm
> [DCM] 20 10 SH 2 2694 StudyID : [ 1 ]
> [DCM] 20 11 IS 2 2704 SeriesNumber : [ 4 ]
> [DCM] 20 12 IS 4 2714 AcquisitionNumber : [ 258 ]
> [DCM] 20 13 IS 2 2726 InstanceNumber : [ 3 ]
>
> 0004.dcm
> [DCM] 20 10 SH 2 2690 StudyID : [ 1 ]
> [DCM] 20 11 IS 2 2700 SeriesNumber : [ 4 ]
> [DCM] 20 12 IS 2 2710 AcquisitionNumber : [ 3 ]
> [DCM] 20 13 IS 2 2720 InstanceNumber : [ 4 ]
>
> The data is public domain (Pittsburgh brain competition 2009 [1]). I
> can upload the data somewhere if you like.
>
> [1]: http://pbc.lrdc.pitt.edu/?q=2009b-download
>
> -F
>
> On Mon, Jul 23, 2012 at 7:32 PM, Donald Tournier
> <d.tournier at brain.org.au> wrote:
> > Hi Félix,
> >
> > I agree that doesn't sound too good. I don't understand how this could
> > happen, unless the series numbers stored in the DICOM header for each
> file
> > are also in the wrong order - this is what MRtrix uses to order the
> volumes.
> > My first guess would be that this is the problem. You can easily check by
> > running read_dicom on each file to see what series number it reports, and
> > whether they match the order you expected. If not, I'd need to take a
> look
> > at the data - any chance you could send me these files?
> >
> > Cheers,
> >
> > Donald.
> >
> >
> > On 24 July 2012 03:57, Félix C. Morency <felix.morency at gmail.com> wrote:
> >>
> >> Hi,
> >>
> >> I am trying to create a 4D Nifti volume from multiple Siemens DICOM
> >> mosaic. The XYZ axis seem to be fine. However, slices along the time
> >> axis are presented in random (?) order.
> >>
> >> I reproduced the issue with four original DICOM mosaic: 000[1-4].dcm.
> >> I convert them to 4D Nifti using mrconvert. The resulting time axis is
> >> ordered as following: 0001, 0004, 0002, 0003, which is not the
> >> expected behaviour.
> >>
> >> Do you have any idea what's going on?
> >>
> >> Regards,
> >> --
> >> Félix C. Morency, M.Sc.
> >> Plateforme d’analyse et de visualisation d’images
> >> Centre Hospitalier Universitaire de Sherbrooke
> >> Centre de recherche clinique Étienne-Le Bel
> >> Local Z5-3031 | 819.346.1110 ext 16634
> >> _______________________________________________
> >> Mrtrix-discussion mailing list
> >> Mrtrix-discussion at www.nitrc.org
> >> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
> >
> >
> >
> >
> > --
> > Jacques-Donald Tournier (PhD)
> > Brain Research Institute, Melbourne, Australia
> > Tel: +61 (0)3 9035 7033
>
>
>
> --
> --
> Félix C. Morency, M.Sc.
> Plateforme d’analyse et de visualisation d’images
> Centre Hospitalier Universitaire de Sherbrooke
> Centre de recherche clinique Étienne-Le Bel
> Local Z5-3031 | 819.346.1110 ext 16634
>



-- 
Jacques-Donald Tournier (PhD)
Brain Research Institute, Melbourne, Australia
Tel: +61 (0)3 9035 7033
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20120725/f270941e/attachment.html