> I have some DICOM images to convert. I have used several packages, but I
> was wondering how the spm_DICOM m-file is doing this. More specifically, I
> am looking how to relate the output of the SPM native conversion (header
> and matlab file) to the differences in origin between the formats. From the
> matlab file I gather that the spm functions flip the DICOM correctly along
> the Y-axis, and that after normalisation in T&T space the images are
> flipped again along the X-axis. Is this correct?
SPM2 does not flip data during spatial normalisation. In SPM99, there used to
be an optional flip. See
http://www.fil.ion.ucl.ac.uk/spm/software/spm2/#Compatability for more info.
>
> Additionally, the matlab file that is output by the SPM conversion, does it
> contain information regarding this transformation?
The .mat file contains a voxel-to-world mapping, where the "world space"
refers to the co-ordinate system within the scanner (but not the same
coordinate system as used by DICOM).
>
> And finally, is the origin determined from the DICOM header and where can
> one find this information.
I would suggest looking at spm_dicom_convert.m. The most relevant lines of
code are probably in the write_volume subfunction, which begins around line
403. For more information about what the DICOM fields mean, check out the
DICOM standard at http://medical.nema.org/dicom/2004.html .
> I have been looking at different sites and manuals but by now I am
> completely lost in the differences between image formats. Any help would be
> greatly appreciated. Moreover, the information about the SPM conversion is
> very hard to find (for instance, the WIKI refers to SPM_TBR, which in spm2
> is obsolete I guess).
spm_TBR is a local thing that we use at the FIL/ICN. The TBR stands for
Trajectory Based Reconstruction, and it is how we generate our Siemens EPI
images from K-space data, making use of the extra knowledge about the
trajectory through K-space.
Best regards,
-John
|