[FieldTrip] ft_volumerealign

[Lilla.Magyari at mpi.nl]

hi Markus,

When you use the ft_volumerealign function it doesn't necessarily will
show the right side of the brain on the right side of the image.
Orientation of the image depends on the coordinate system of the mri you
read in.

 I tried also align the template mri to the fiducials, and I got the same
problem as you (z+ going to inferior in the aligned volume), but when I
switched the lpa with the rpa the orientation was fine. So, my conclusion
is that the right side shows up on the left of the image when you use
ft_volumerealign. Therefore, you have to mark the lpa on the right side
of the image and the rpa on the left side of the image. But there is also
another option implemented exactly because of this left/right flipping:
You can determine also a "z-point" (by pressing z in the image) that can
be anywhere at the top of the head. Then, the function will give you the
right alignment even if  you did not determine the lpa and rpa on the
right side of the brain.
For this, you just have to do this:
cfg.fiducial.zpoint  = [i j k], position of zpoint
when you also determine  the positions for the nasion, lpa and rpa.

And one more remark: I do not know if you have realized this yourself, and
I have just misunderstood your email: When you call ft_volumerealign in
interactive mode like this:

cfg=[];
cfg.method='interactive';
mri_realigned=ft_volumerealign(cfg,mri);

And you press the keys r/l/n/z at the right places in the volume (and quit
with q), the output (mri_realigned) will be the already aligned volume,
you do not need to call ft_volumerealign again with method 'fiducials'.

Best,
Lilla

> Dear list
>
> when running ft_volumerealign with the spm template MRI in the interactive
> mode, I get something like: 
>
>
> ===========================================================================voxel
> 838065, indices [46 54 85], location [0.0 -20.0 96.0] mm
> nas_voxel = [46.000000 104.000000 11.000000], nas_head = [0.000000
> 80.000000 -52.000000]
> lpa_voxel = [5.000000 54.000000 11.000000], lpa_head = [82.000000
> -20.000000 -52.000000]
> rpa_voxel = [85.000000 54.000000 11.000000], rpa_head = [-78.000000
> -20.000000 -52.000000]
> xzpoint_voxel = [46.000000 54.000000 85.000000], xzpoint_head = [0.000000
> -20.000000 96.000000]
> the call to "ft_volumerealign" took 136 seconds and an estimated 39 MB
>
>
> Now I would like to rerun it in the fiducial mode, using the above
> information.
> According to help, I need to do this:
>
>
> For realigning to the fiducials, you should specify the position of
> the fiducials in voxel indices.
>     cfg.fiducial.nas  = [i j k], position of nasion
>     cfg.fiducial.lpa  = [i j k], position of LPA
>     cfg.fiducial.rpa  = [i j k], position of RPA
>
>
>
> I used the voxel indices (code below), but ended up with the MRI
> upside-down (see attached picture).
> Trying the landmark mode was also without success.
>
>
> Am I doing something silly here?
>
>
>
> Thanks in advance
> Markus
>
>
>
>
> % Load SPM8 T1 template 
>
>
>  template     = '/Documents/MATLAB/spm8/templates/T1.nii';
>
>
>  template_mri = ft_read_mri(template);
>
>
>  disp(template_mri)
>
>
>  
>
>
>  % 2. Realign the coordinate system
>
>
>  cfg                = []; 
>
>
>  cfg.method         = 'fiducial'; % 'interactive';
>
>
>  cfg.fiducial.nas   = [46 104 11];
>
>
>  cfg.fiducial.lpa   = [5 54 11];
>
>
>  cfg.fiducial.rpa   = [85 54 12];
>
>
>  cfg.fiducial.zpoint= [46 54 85];
>
>
>  cfg.coordsys       = 'ctf';
>
>
>  template_mri_realigned  = ft_volumerealign(cfg, template_mri);
>
>
>
>
>
>
>  
>
>
>  % 3. Check the coordinate system
>
>
>  cfg                     = [];
>
>
>  cfg.interactive         = 'yes';
>
>
>  template_mri_checked    = ft_determine_coordsys(template_mri_realigned);
>
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip