[MINC-users] dcm2mnc vs 3D DICOM
Alex Zijdenbos
zijdenbos at gmail.com
Mon Feb 15 12:29:34 EST 2016
Hi Bert,
I have admit I don't have a strong preference one way or the other. If the
DICOM reports a scanner-world-coordinate origin (0,0,0), it probably means
that the 'real' coordinate information was lost or never existed (e.g.,
data was processed). In that case it probably doesn't matter as any further
processing would need explicit registration anyways. I can imagine that,
when unknown, placing the world coordinate origin in the center of the
image, as dcm2nii appears to do, is a practical choice, but not really
better than using (0,0,0) that the DICOM actually reports. Only advantage I
can think of is that it brings native volumes up seeming better aligned
with other modalities. It's pretty trivial to shift the starts to another
location after the fact anyways.
I don't know if anybody else has opinions about this, but AFAIC we can
leave this as-is.
-- A
On Thu, Feb 11, 2016 at 4:11 PM, Robert D. Vincent <
robert.d.vincent at mcgill.ca> wrote:
> Hi Alex,
>
> I'll try to double check our handling of this. It could be that some other
> field in the DICOM is instructing that the position should be treated as a
> centre rather than an corner. But my understanding has always been that the
> position is relative to the upper left hand corner of the image.
>
> It's likely that the dcm2nii developers implemented a kludge for a zero
> position. If you think it's worth it, we could certainly do the same.
>
> -bert
>
> On Thu, Feb 11, 2016 at 3:59 PM, Alex Zijdenbos <zijdenbos at gmail.com>
> wrote:
>
> > Hi Bert,
> >
> > Yes - converting to nii and then to MINC yields this volume:
> >
> > image: signed__ short 0 to 32766
> > image dimensions: zspace yspace xspace
> > dimension name length step start
> > -------------- ------ ---- -----
> > zspace 65 3.138 -101.985
> > yspace 128 2.353 -150.592
> > xspace 128 2.353 -150.592
> >
> > So I suspect you are right... Either way is fine with me in practice, I
> > just wanted to make sure the conversion wasn't dropping information.
> >
> > Thanks!
> >
> > -- A
> >
> >
> > On Thu, Feb 11, 2016 at 3:43 PM, Robert D. Vincent <
> > robert.d.vincent at mcgill.ca> wrote:
> >
> > > Hi Alex,
> > >
> > > That is surprising. There is only one position indicated in the file,
> and
> > > we are correctly reading it as 0, 0, 0 (it's a DICOM element 0020,
> 0032).
> > > So as far as we are concerned we're assigning the correct start values.
> > >
> > > It's possible that dcm2nii is "inventing" position values corresponding
> > to
> > > the centre of the grid - are they something like 150.6? (128/2)*2.353
> > >
> > > -bert
> > >
> > >
> > > On Thu, Feb 11, 2016 at 3:02 PM, Alex Zijdenbos <zijdenbos at gmail.com>
> > > wrote:
> > >
> > > > Hi Bert,
> > > >
> > > > Cool - thanks much! The steps/zspace now look good :-)
> > > >
> > > > However I believe that the starts may still be wrong. When run
> through
> > > > dcm2nii - nii2mnc, the volume ends up with start values that I
> suspect
> > > are
> > > > 'real' (they roughly match the corresponding T1 that I have). With
> this
> > > new
> > > > dcm2mnc, the start values are all 0.
> > > >
> > > > -- A
> > > >
> > > > On Thu, Feb 11, 2016 at 2:34 PM, Robert D. Vincent <
> > > > robert.d.vincent at mcgill.ca> wrote:
> > > >
> > > > > Hi Alex,
> > > > >
> > > > > I just pushed a fix to Github that should put correct Z axis
> spacing
> > in
> > > > > these sequences.
> > > > >
> > > > > -bert
> > > > >
> > > > > On Thu, Feb 11, 2016 at 10:59 AM, Gabriel A. Devenyi <
> > > gdevenyi at gmail.com
> > > > >
> > > > > wrote:
> > > > >
> > > > > > Just a follwup, the "newer" dicom creator is dcm2niix, may be
> able
> > to
> > > > > find
> > > > > > some useful info here https://github.com/neurolabusc/dcm2niix
> > > > > >
> > > > > > --
> > > > > > Gabriel A. Devenyi B.Eng. Ph.D.
> > > > > > Research Computing Associate
> > > > > > Computational Brain Anatomy Laboratory
> > > > > > Cerebral Imaging Center
> > > > > > Douglas Mental Health University Institute
> > > > > > McGill University
> > > > > > t: 514.761.6131x4781
> > > > > > e: gdevenyi at gmail.com
> > > > > >
> > > > > > On Thu, Feb 11, 2016 at 10:47 AM, Robert D. Vincent <
> > > > > > robert.d.vincent at mcgill.ca> wrote:
> > > > > >
> > > > > > > Hi Alex,
> > > > > > >
> > > > > > > We've encountered so little 3D DICOM over the years I'm not
> > > surprised
> > > > > it
> > > > > > > isn't handled well.
> > > > > > >
> > > > > > > Thanks for the dataset. I will look into it and let you know.
> > > > > > >
> > > > > > > -bert
> > > > > > >
> > > > > > >
> > > > > > > On Thu, Feb 11, 2016 at 10:33 AM, Alex Zijdenbos <
> > > > zijdenbos at gmail.com>
> > > > > > > wrote:
> > > > > > >
> > > > > > > > Hi all,
> > > > > > > >
> > > > > > > > I got some 3D DICOM SPECT data that dcm2mnc issues with -
> > > depending
> > > > > on
> > > > > > > what
> > > > > > > > version of dcm2mnc I use, I get somewhat different results;
> but
> > > in
> > > > > most
> > > > > > > > cases I end up with irregular z-spacing and/or a flipped
> > z-axis.
> > > > The
> > > > > > most
> > > > > > > > recent version I tried was dcm2mnc version 2.01.03 from
> > > > minc-toolkit
> > > > > > > > version 1.0.08.
> > > > > > > >
> > > > > > > > If I run the same thing through dcm2nii followed by nii2mnc
> > > > (ugh!), I
> > > > > > > > actually get a volume that looks like I expected. There
> appears
> > > to
> > > > be
> > > > > > > > coordinate space information in the 3D DICOM that dcm2nii
> seems
> > > to
> > > > be
> > > > > > > able
> > > > > > > > to interpret, but dcm2mnc does not.
> > > > > > > >
> > > > > > > > Thoughts/pointers/updated versions of dcm2mnc?
> > > > > > > >
> > > > > > > > Here is a sample DICOM set:
> > > > > > > >
> > > > > > > > s1_0013006_3006.dcm.gz
> > > > > > > > <
> > > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
> https://drive.google.com/file/d/0B5fOtqpIs4sKVVlVVVF0VTZ5Sjg/view?usp=drive_web
> > > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > > > Thanks,
> > > > > > > >
> > > > > > > > -- A
> > > > > > > > _______________________________________________
> > > > > > > > MINC-users at bic.mni.mcgill.ca
> > > > > > > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > > > > > > >
> > > > > > > _______________________________________________
> > > > > > > MINC-users at bic.mni.mcgill.ca
> > > > > > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > > > > > >
> > > > > > _______________________________________________
> > > > > > MINC-users at bic.mni.mcgill.ca
> > > > > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > > > > >
> > > > > _______________________________________________
> > > > > MINC-users at bic.mni.mcgill.ca
> > > > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > > > >
> > > > _______________________________________________
> > > > MINC-users at bic.mni.mcgill.ca
> > > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > > >
> > > _______________________________________________
> > > MINC-users at bic.mni.mcgill.ca
> > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > >
> > _______________________________________________
> > MINC-users at bic.mni.mcgill.ca
> > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> >
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>
More information about the MINC-users
mailing list