Hi Alle Meije,
I've used SPM5's DICOM import and dinifti with a bunch of unsorted
DICOM slices (not a DICOMDIR, but I think it should be similar) rather
than volumes, and they seemed to work fine (though as I mentioned some
DICOM data ended up with different NIfTI qforms depending on the
program -- something I keep meaning to look into more carefully...)
One trick that might help though (again, I haven't looked into this)
is to use medcon to stack the DICOM slices into a DICOM volume, and
then to convert this DICOM volume to a NIfTI one with e.g. dcm2nii etc.
At one point I was using a simple command to create a DICOM volume
medcon -f `slice_set.pl` -c dicom -stack3d
having written a slightly more complicated Perl script, slice_set.pl,
to pick out and order the slices (for some rather messed up
"anonymised" DICOM data that I got, which confused all the automated
series-ordering software). Also, if you haven't already spotted them,
the "mklinks" scripts from the medcon FAQ are very useful
http://xmedcon.sourceforge.net/Faq/Stack
http://xmedcon.sourceforge.net/pub/mklinks
http://xmedcon.sourceforge.net/pub/mklinks2
You'd think that getting the data from the scanner to a form that
could be analysed easily with FSL/SPM/etc, while preserving as much
information as possible, would be a solved problem by now, wouldn't
you? ;-)
Best of luck,
Ged
>> Going from DICOM to NIfTI via Analyze like this loses the orientation
>> information in the DICOM fields (e.g. ImagePatientPosition, etc)
>> [snip]
>> Some of the direct DICOM->NIfTI programs set the NiFTI qform from the
>> DICOM fields (although, when I checked this, a couple of them were
>> inconsistent, so do take care!) I think these programs try this:
>> - dcm2nii
>> - SPM5 DICOM import
>> - dinifti
>> - possibly LONI Debabaler and others
>> Links to all these are at the bottom of Chris Rorden's dcm2nii page:
>> http://www.sph.sc.edu/comd/rorden/mricron/dcm2nii.html
>
> I have tried a few of these, but most of them seem to expect volume/4D
> files to begin with or simply didn't work. In the DICOMDIR setting,
> scripting it together and having medcon make the volumes, was the best
> way I could convert whole batches of files from a CD.
>
> Plus, are you sure that these programs estimate the position of a volume
> using the positions of individual slices?
>
> I tried the links, but none of them actually gave useful information in
> terms of: this is your DICOM cd, and this is how you rip your volumes
> off it...
>
> Any advice is still much appreciated, and if those programs really come
> up with Q-forms from slice data, I'm very impressed!
|