[FieldTrip] rpa/nas/lpa coordinates

Robert Oostenveld r.oostenveld at donders.ru.nl
Mon May 23 09:33:34 CEST 2011

Dear Elena

> On 21 May 2011, at 14:11, Elena Orekhova wrote:
>> Hi
>> I Just started to use Fieldtrip with my Neuromag MEG data and try  
>> to follow  the example script
>> ‘read_neuromag_mri_and_create_single- 
>> subject_grids_in_individual_head_space_that_are_all_aligned_in_mni_space 
I don't know all details of the Neuromag system, but recently we did  
add the http://fieldtrip.fcdonders.nl/getting_started/yokogawa  
documentation that explains reasonably general how coordinate system  
coregistration works. We also added a new function (FT_SENSORREALIGN)  
to ease the coregistration and we are working towards a more  
consistent representation of the corodinate systems in the relevant  
fieldtrip data structures.

Although the details for your system are different (see http://fieldtrip.fcdonders.nl/faq/how_are_the_different_head_and_mri_coordinate_systems_defined) 
, you might get a better overview of the steps that are involved by  
reading the yokogawa getting started.

> The instruction suggests to take voxel coordinates of fiduscials  
> from the Neuromag
> GUI for MRI-MEG Integration. I have already performed this step and  
> saved the
> transformed mri in the /sets/XXX.fif file.

I suggest you use the FT_DETERMINE_COORDSYS to perform a visual  
inspection of the coordinate systems of your anatomical MRI and of  
your sensor positions (data.grad, or you can use FT_READ_SENS to read  
them directly from the MEG fif file). Note that you have to use the  
function twice (on the mri and on the grad) and compare the figures  

>  Should I write any  rpa/nas/lpa coordinates in this case?
> cfg.fiducial.rpa  = [y x z];
> cfg.fiducial.nas  = [y x z];
> cfg.fiducial.lpa  = [y x z];

if you need them (see above), then these should be [x y z] and not [y  
x z].

> I have 2 4X4 transformations in the .fif file, not sure what they  
> mean.

Please read this page, which is a general introduction.

> mri.hdr.trans.trans
> 0.9913    0.1210    0.0518   -0.0032
>    -0.0948    0.9291   -0.3574    0.0254
>    -0.0914    0.3494    0.9325   -0.0519
>          0         0         0    1.0000
> mri.hdr.voxel_trans.trans
>    -0.0000   -0.0000   -0.0010    0.1058
>    -0.0005    0.0000    0.0000    0.1597
>          0   -0.0005    0.0001    0.1058
>          0         0         0    1.0000

the mri header is system specific, so I am not sure. But I guess that  
the first is the homogenous transformation matrix in head coordinates  
(linked to the fiducials) and the second is the homogenous  
transformation matrix in MRI coordinates. The first one appears to be  
in mm, the second one in meter.

Please try

mri.transform = mri.hdr.voxel_trans.trans;
ft_determine_coordsys(mri)  % press ctrl-c

mri.transform = mri.hdr.trans.trans;
ft_determine_coordsys(mri)  % press ctrl-c

The two figures will show you the origin, the direction of axes of the  
coordinate system and the scale.

best regards,

More information about the fieldtrip mailing list