[MINC-users] dcm2mnc produces xspace:spacing = "irregular"

Steve M. Robbins steve at sumost.ca
Mon Apr 21 12:05:13 EDT 2008


On Tue, Apr 22, 2008 at 01:12:08AM +1000, Andrew Janke wrote:
> >  > Unfortunately DICOM does not always unambiguously describe the
> >  > position of the slices in an image volume (some scanners use
> >  > origin/angle, some use absolute position of centre of slices, some use
> >  > offsets, etc).
> >
> >  What modalities are you talking about?  I grant you that NM does weird
> >  things (possibly US, too).  However for MR, CT, and PET, my reading of
> >  DICOM suggests they all use the same (0020,0032) Image Position
> >  (Patient) and (0020,0037) Image Orientation (Patient) attributes to
> >  specify the location and orientation of the slices.
> >
> >  What ambiguity are you referring to?
> 
> That is correct they are _supposed_ to put things in the right
> places... To put the first spanner in the works, consider the special
> case of "mosaic" images.  I figure they came about to get around some
> of the issues of bazillions of files in a directory for functional
> data.

Sure.  But your problem is not with DICOM per se, but rather with
interpreting the vendor private attributes that presumably define the
slice positions.  I've never seen a mosaic image; I'd be interested to
have a look if you have an example handy.

You could rightly argue the problem stems from DICOM not having a
useful 3D or 4D format when fMRI was developed.  Now that they have
the Enhanced MR IOD, is anyone using it for fMRI?


> Then we will shift on to how some scanners define a slice gap as being
> part of a slice thickness and some dont.  Ah, where would we be
> without diversity? :)

Not sure how this is relevant to the position.  Slice position is
unambiguously defined using (0020,0032).  Slice Thickness is an
optional attribute and useless (for location) even when present
because thickness is completely independent of slice-to-slice spacing.
I've never heard of a Slice Gap attribute.


Both these examples, however, are a far cry from "some scanners use
origin/angle, some use absolute position of centre of slices, some use
offsets, etc", which is what I was curious about.

Chimo,
-Steve


More information about the MINC-users mailing list