Dear Wibeken,
it could happen that voxel size of converted images was wrong due to
using a different matrix for acquisition and reconstruction. Voxel
size stored in dicom header could be, depending on the manufacturer,
both the acquisition or the reconstruction one (one is half size the
other in x and y axis). You could check if this is your problem with
spm's DISPLAY tool. If brains look deformed, or the total brain size
is too large (number of voxels multiplied by the voxel size), you
could fix the problem by resizing your images by a factor 0.5
(probably just along x and y axis).
Hope this helps
Carles
Wibeke Nordhøy <[log in to unmask]> ha escrit:
> Hello SPM-users:
>
> We've been trying to calculate B0 maps based on dual gradient echo
> images from our GE Signa scanner. We used the GRE sequence, with a
> 128 acquisition matrix reconstructed to a 256 matrix, making Re and
> Im images for the two echo times (4.9 ms and 7.4 ms). Then we
> converted these to NIfTI using either MRIcro or SPM8 before running
> the Fieldmaps toolbox on them. After a while, the analysis will
> crash and produce the following errors:
>
> ??? Error using ==> mrdivide
> Matrix dimensions must agree.
>
> Error in ==> pm_smooth_phasemap at 49
> pm = pm_smooth_phasemap_dtj(pm,1/angvar,krnl);
>
> Error in ==> pm_make_fieldmap at 201
> fm.fpm = pm_smooth_phasemap(fm.upm.*fm.mask,angvar,pxs,flags.fwhm);
>
> Error in ==> FieldMap at 1623
> IP.fm = pm_make_fieldmap([IP.P{1} IP.P{2} IP.P{3}
> IP.P{4}],IP.uflags);
>
> Error in ==> FieldMap at 699
> status=FieldMap('CreateFieldMap',IP);
>
> ??? Error while evaluating uicontrol Callback
>
> It seems that the Fieldmaps tool is geared towards Siemens scanners;
> have anyone used it on images from a GE scanner with good results?
> Any tips or ideas for us?
>
> The first error message mentions matrix dimensions. Could our
> problem be the difference between the acquired and reconstructed
> matrices?
>
> Wibeke and Øystein
>
|