Re: [HCP-Users] where are the head shape file for MEG

[Glasser, Matthew]

Generally I thought you could only directly subsample the regular icosahedron spheres (e.g. 2562, 10242, 40962, 163492).  The native mesh sphere isn’t going to be amenable to that sort of thing as it is very irregular (varying numbers of neighbors per vertex, varying vertex spacing, etc).  One can certainly resample from one spherical mesh to another (this is how we get MRI data onto a standard mesh were vertices have correspondence across subjects).  Additionally, one can resample an anatomical surface from one mesh to another.  When this is done, the surface remains in the same volume space (could be native volume space, MNI volume space, FreeSurfer coordinate space, even “helmet space” if that is not the same as one of the previous 3 spaces).  I’m guessing that there is not any theoretical reason why any of the MEG software cannot use standard meshes, but perhaps there are implementation issues. Peace, Matt. From: Denis-Alexander Engemann <denis.engem...@gmail.com> Date: Monday, May 18, 2015 at 11:56 AM To: Matt Glasser <glass...@wusm.wustl.edu> Cc: jan-mathijs schoffelen <jm.schoffe...@gmail.com>, Jair Montoya <montoya.j...@gmail.com>, "Elizabeth Anne Bock, Ms" <elizabeth.b...@mcgill.ca>, Alexandre Gramfort <alexandre.gramf...@gmail.com>, "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org>, Eric Larson <larson.eri...@gmail.com> Subject: Re: [HCP-Users] where are the head shape file for MEG Hi, I included Eric here who also is an author of PySurfer which we intensely use for visualization. In MNE you subsample the original surfaces to define dipole grids. For subsequent morphing and plotting operations you need to know the slection of original vertices used. If you store a source model, you normally would store all triangles and positions of the original tessalation, the reduced ones, the surface normals, and mapping indices to connect the subsampled locations to the original ones. Also the MNE inverse pipeline rely on the freesurfer coordinate system. To deal with this latter point you would at least need know how to transform the MNE source models to the helmet space ... Best, Denis 2015-05-18 18:42 GMT+02:00 Glasser, Matthew <glass...@wusm.wustl.edu>: Can someone explain why it is important to use original FreeSurfer surfaces?  These have a much higher density that is needed for most applications, are very irregular, and do not have vertex correspondence across subjects.  It is possible to resample back to the FreeSurfer native mesh, but I am puzzled as to why that would be necessary. Peace, Matt. From: Denis-Alexander Engemann <denis.engem...@gmail.com> Date: Monday, May 18, 2015 at 9:35 AM To: jan-mathijs schoffelen <jm.schoffe...@gmail.com> Cc: Jair Montoya <montoya.j...@gmail.com>, "Elizabeth Anne Bock, Ms" <elizabeth.b...@mcgill.ca>, Alexandre Gramfort <alexandre.gramf...@gmail.com>, "hcp-users@humanconnectome.org" <hcp-users@humanconnectome.org> Subject: Re: [HCP-Users] where are the head shape file for MEG > However, we used wb_command for the subsampling, which as far as I know is in fact doing a re-tessellation on the sphere, rather than selecting a subset of vertices.  Arff. My prediction is that computing solutions will ultimately work, all plotting will be severly broken as this involves the original freesurfer surfaces. It would be good to have a mapping between the resampled vertices and the original vertices if this is possible at all (?). Any chances we can trigger some workaround / action to get the trafos between bti and freesurfer? This would facilitate many things and would allow for a more complete bridge between the HCP MEG data and different software environments. Where would you start? Best, Denis 2015-05-18 7:30 GMT+02:00 jan-mathijs schoffelen <jm.schoffe...@gmail.com>: Hi Denis, Indeed the meshes’ vertices are indeed already registered in surface space, i.e. to the fsaverage, and are obtained from the 32k meshes provided in the structural processing package. However, we used wb_command for the subsampling, which as far as I know is in fact doing a re-tessellation on the sphere, rather than selecting a subset of vertices. This means that it’s probably not straightforward to generate an ‘inuse’ vector. Best, Jan-Mathijs On May 18, 2015, at 4:04 AM, Denis-Alexander Engemann <denis.engem...@gmail.com> wrote: ... taking a closer look at the reference manual it turns out the dipoles are already in the standard model space but morphed to individual anatomy, at least for the 3d models. I'll take a look at `hcp_anatomy.m` to better understand the details that would facilitate representing the source models appropriately in MNE data containers which normally contain a few additional variables (mostly related to the subsampling of the full freesurfer output) and some meta info. Best, Denis   2015-05-18 2:25 GMT+02:00 Denis-Alexander Engemann <denis.engem...@gmail.com>: Hi Jan Mathijs, the reason was that we're aiming for HCP I/O support in MNE that is as complete as possible. This means it would be nice to be able to use MNE functions to create source models which are defined in freesurfer space. What you suggest might indeed work, I'm giving it a try. That would already be a start. But I fear it will not be easy to make some standard MNE routines work with the shipped source models, , e.g, morphing to fsaverage, etc. This is mainly because of the coordinate system. But maybe there is a solution to this in the HCP-MEG pipeline we could learn from? As always any hints are welcome. Best, Denis 2015-05-15 22:09 GMT+02:00 jan-mathijs schoffelen <jm.schoffe...@gmail.com>: Hi Denis, Following up on my previous comment: why is it exactly you want the mapping to freesurfer space? The 123456_meg_anatomy_sourcemodel_2d.mat is already in helmet space. If you want to use it with MNE suite to create leadfields that are expressed in the correct coordinate system, I think writing the contents of the mat-file mentioned to a MNE-suite compatible file with an identity matrix as the transform. Or am I missing something? Best, Jan-Mathijs On May 15, 2015, at 3:44 PM, Denis-Alexander Engemann <denis.engem...@gmail.com> wrote: I was wondering wheather the transforms would be suitable for mapping the freesurfer surface coordinates to the helmet space. If this step would work, the transforms and the coreg would be usable with other software such as for example the MNE suite. We could then write a routine that skips the regular conversion of BTI files and adds identity matrices instead to the related transform fields that usually map from BTI to Neuromag space. The transforms provided could then be used to map the sensors to the source models created in MNE. Any hint wold be appreciated.  Best, Denis   2015-05-15 15:35 GMT+02:00 Georgios Michalareas <giorgos.michalar...@esi-frankfurt.de>: Dear Beth, the head shape files are not provided with the data. The reason for this is that headshape digitization includes points around the nose, the eyes, the forehead and the entire head. This means that this information could lead to identification of a subjects identity. This is also the reason why the structural MR images provided by the MR modality have also such external features stripped off the data in order to avoid a possible subject identification. Following this constraint, in the MEG team we provide both surface(2D) and volumetric(3D) source models so that the community can perform source localization on them. You can find them in the anatomy directory of the released data. The surface source model is called 177746_MEG_anatomy_sourcemodel_2d.mat     while there are  3 volumetric models with different voxel sizes (4,6,8 mm) 177746_MEG_anatomy_sourcemodel_3d4mm.mat  177746_MEG_anatomy_sourcemodel_3d6mm.mat  177746_MEG_anatomy_sourcemodel_3d8mm.mat  I hope this helps. Let me know for anything else. Giorgos On 15/05/2015 15:18, Elizabeth Anne Bock, Ms wrote: I just downloaded the MEG data from the 500 Subjects + MEG2 Data Release.  I see from the documentation that the MEG data should include a head shape file.  This is not present in my downloaded set.  Where can I get these? Thanks, Beth ------------------------------------------------------------------------------------------ Elizabeth Bock / MEG System Engineer McConnell Brain Imaging Centre / Montreal Neurological Institute McGill University / 3801 University St. / Montreal, QC H3A 2B4 MEG Lab: 514.398.6056 Mobile: 514.718.6342 _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users This email is free from viruses and malware because avast! Antivirus protection is active. _______________________________________________ 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 _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users   The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.   The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail. _______________________________________________ HCP-Users mailing list HCP-Users@humanconnectome.org http://lists.humanconnectome.org/mailman/listinfo/hcp-users