[MINC-users] DICOM (MOSAIC) to MINC conversion

[Andrew Janke]

On Wed, 21 Jul 2004, Roy Yves wrote:

> 1- Is there a DICOM (MOSAIC) to MINC converters available somewhere? Or do I
> have to do the conversion DICOM -> ANALYSE -> MINC (which doesn't appear to be
> a resonnable solution to me ==> See question 5 below).
>
> ==> I am about to look at and evaluate dicom3_to_minc and LONI Debabeler...
> (and I will report my results on the REPRIC web site, conversion tools page
> http://www.rsmnq.ca/repric/en/conversion_software.htm, and your comments are
> also more than welcome).

As far as I am aware neither dicom_to_minc or dicom3_to_minc will help you here.

However if you happen to have Siemens (it will work with no other mosaic files!)
mosaic files then I have hacked together something that reads the SIEMENS
"ASC_CONV" header within the DICOM header and uses this to convert the data.

   http://www.cmr.uq.edu.au/~rotor/software/progs/siemens_mosaic2mnc

I'd be 'interested' if it works for you. :)

> 2- On one of the BIC's machine, I simply ran the installed dicom_to_minc on a
> (supposedly DICOM-MOSAIC, ==> See question 4 below) image to obtain an output
> MINC image of dimensions 384 x 384 x 1 (as given by mincinfo) pretty much all
> black (the conversion didn't work, --but I did not get any error messages...).
> How can these numbers make sense knowing that the original DICOM image were
> (supposedly) (64 x 64) x 28 slices ?

384 = 64 x 6.

the "mosaic" format is a very literal mosiac of images.  I suspect the ouput of
dicom3_to_minc will be a nice little array of 6x6 images in each 'slice'.

Take a look at the guts of siemens_mosaic2mnc it is little more than a wrapper
that then chops the resulting converted minc file up into bits.

> 3- Is the *new* dicom3_to_minc supporting the conversion from DICOM-mosaic to
> MINC ?

Not that I know of.  Mosaic format is eseentially yet another hack perpetrated
upon us due to the limitations of DICOM (2D files only) and since fMRI makes
lots of them a few files are shoe-horned into one.

> 4- Given an image (which is assumed to be) in DICOM format (say the info about
> the origin was lost...):
>
> - How can the true format of the image be found (the flavor, DICOMDIR or
> MOSAIC, the dimensions, and other reflective info like that), i.e. what are
> the available tools allowing that?

This ofter depends on the source of the DICOM image :(.  I doubt DICOMDIR will
help you here, it typically is a catalog of the files in the DICOM directory
with a bit of header info included (scan type, sequence#, patient etc).

> - I read somewhere (http://research.bmap.ucla.edu/MRIFAQ.html) that the latest
> version of SPM is reportedly able to handle the mosaic format. Anyone knowing
> whether it is true? Would it be a mosaicised *Analyse* format or *DICOM*
> supported (DICOM does not work for me under SPM2...) ?

There is no such thing as ANALYZE mosaic that I know of. (please no!) With
respect to SPM reading mosaic, I suspect it can via some plug-in.  The other
option is of course xmedcon.

> I got the following error in step 3:

> PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0,
> 'blib/lib', 'blib/arch')" t/*.t
> make: *** [test_dynamic] Segmentation fault

wierd, which platform?

However this may be somewhat benign, does the make install step work?


--
Andrew Janke  ( rotorATcmr.uq.edu.au || www.cmr.uq.edu.au/~rotor )
Australia->University of Queensland->Centre for Magnetic Resonance
W: +61 7 3232 7254  ||  H: +61 7 3800 4042  ||  M: +61 4 2138 8581