Hi Tzvetan, It sounds you want to do a beamformer type of source reconstruction using source positions defined on a 3D grid. You could use the provided data for this, where the individual grids are in register with a grid in normalised volume. Assuming you are doing all in FieldTrip anyway, you could interpolate the volumetric atlas onto one of the template grids (which are part of the software release, but can also be found in FieldTrip), which will give you the anatomical labelling for each of your beamformer’s dipole positions right away. Alternatively, if you would really want to create your own 3D-grid, you should plug in the appropriate transformation matrix from the transform text-file into the anatomical MR-structure. A different route altogether, and this is what the HC-consortium is pushing for, is to work in surface-registered space. This does much more credit to the analysis of the data from the haemodynamic and structural modalities. Although it’s still an open question whether MEG in itself will benefit from a surface-based analysis (when using beamformers), you could consider using the cortical sheet source models, in combination with an anatomical labeling defined on the surface (e.g. the Conte69 atlas). Of note, if you were working on the surface, you could benefit from any cool parcellation schemes that may be coming out of the consortium soon.
Best wishes, Jan-Mathijs On Apr 27, 2015, at 9:37 AM, Tzvetan Popov <tzvetan.po...@gmail.com> wrote: > > Hi Jan-Mathijs, > >> Hi Tzvetan, >> >> Which anatomical MR are you using? > I’m using '100307_3T_T1w_MPR1.nii.gz’. >> Note that the ‘baked in’ fiducial locations are indeed approximate, but they >> only make sense in context of the anatomical image in which they have been >> clicked. This image was a 1mm resolution image with a sufficient field of >> view to capture the entire scalp surface + facial details. This image is not >> part of the released data. > Got it. Want work then indeed. I assumed that the approximate locations are > based on the provided MRI. All clear. >> Thus, in short, the fiducial locations (defined in voxels) don’t make sense >> in the context of the anatomical image you are using. >> Since everything is already in register (and the co-registration details are >> represented in the 100307_MEG_anatomy_transform.txt), what is it you want to >> do? > Yes all is co-registred and I was able to do source analysis etc. What I > wanted to do is to re-compute the sourcemodel warped in the aal-mni-space > instead of mni-space as it is now. > Many thanks > tzvetan >> >> Best wishes, >> >> Jan-Mathijs >> >> >> >> Jan-Mathijs Schoffelen, MD PhD >> >> Max Planck Institute for Psycholinguistics, >> Nijmegen, The Netherlands >> >> Donders Institute for Brain, Cognition and Behaviour, >> Centre for Cognitive Neuroimaging, >> Radboud University Nijmegen, The Netherlands >> >> j.schoffe...@donders.ru.nl >> Telephone: +31-24-3614793 >> >> http://www.hettaligebrein.nl >> >> >> >> >> >> >> On Apr 27, 2015, at 8:08 AM, Tzvetan Popov <tzvetan.po...@gmail.com> wrote: >> >>> Dear hcp-users, >>> >>> I am trying to coregister the individual MRI and the 4D coordinates using >>> the fiducial information provided in ‘100307_MEG_anatomy_fiducials.txt’. >>> When using the provided voxel coordinates the anatomical image is >>> misaligned (see attached figure 2). Interactively providing the approximate >>> coordinates of the fiducials aligned the volume correctly (attached figure >>> 1). It appears to me that the provided voxel coordinates are somewhat off. >>> . Am I doing something wrong, has anybody encountered similar problem? >>> Thanks a bunch in advance >>> tzvetan >>> >>> Here is the code snippet that will produce the two figures >>> >>> cfg=[]; >>> cfg.method = 'fiducial'; >>> cfg.fiducial.nas = [ 40 154 94 ];% hcp locations >>> cfg.fiducial.lpa = [ 139 168 169 ];% hcp locations >>> cfg.fiducial.rpa = [ 141 171 26 ];% hcp locations >>> mriR = ft_volumerealign(cfg, mri); >>> ft_determine_coordsys(mriR, 'interactive', 'no’); >>> >>> cfg=[]; >>> cfg.method = 'fiducial'; >>> cfg.fiducial.nas = [128 299 141 ];% my interactive locations >>> cfg.fiducial.lpa = [ 240 148 91 ];% my interactive locations >>> cfg.fiducial.rpa = [ 29 157 80];% my interactive locations >>> mriR = ft_volumerealign(cfg, mri); >>> ft_determine_coordsys(mriR, 'interactive', 'no'); >>> >>> >>> <Misaligned.png><Aligned correctly.png> >>> _______________________________________________ >>> HCP-Users mailing list >>> HCP-Users@humanconnectome.org >>> http://lists.humanconnectome.org/mailman/listinfo/hcp-users >>> >> > _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users
