Dear Jeff,
if I have understood correctly you want to import a structural MRI scan
for a given HCP subject into Brainstorm for creating the EEG source model.
I guess you follow a procedure similar to
http://neuroimage.usc.edu/brainstorm/Tutorials/HCP-MEG
and Brainstorm cannot find the MEG/anatomy/*_MEG_anatomy_transform.txt file.
Although I am not sure whar Brainstorm reads exactly from the transform
file , I can tell you what the transorm file contains. It contains a
matlab structure with some fields, which I think you could easily create
for subject that have no MEG data.
First a quick word on coordinate systems:
spm: This is the MNI coordinate system. This is the coordinate system
that you 0.7 mm processed structural MRI is defined.
bti: This is the coordinate system of the MEG . Of course in your case
where there was no MEG recorded , this will be the same as the "spm"
coordinate system of the structural MRI. If for your EEG you have
digitised the location of the EEG sensors relative to the fiducials then
of course you ll need to figure out the transformation matrix between
the structural MRI and you EEG reference system. But if this is not the
case then your "bti" coordinate system should be the same as the "spm"
coordinate system. In this case the identity transformation matrices
between this coordinate systems would be
spm2bti= [
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
];
bti2spm= spm2bti;
vox: This is the voxel coordinate system of the structural MRI . The MRI
file contains in itself the transformation matrix from voxels to actual
coordinates in the "spm". To get this transformation matric in matlab
you can use the fieldtrip function "ft_read_mri"
Let's assume you have done so and you have found that the transformation
matrix from voxel to coordinates is
vox07mm2spm= [
0.0296775 0.00303441 -0.996087 92.0035
-0.965804 -0.245889 -0.0309473 159.445
0.253714 -0.970409 0.00438009 90.3328
0 0 0 1
];
Important remark: The HCP processed structural MRI has a resolution of
0.7mm. For the MEG processing we use a lower resolution version of the
structural MRI . So in the description below whenever you see the
notation "vox07" refers to the 0.7mm scan and just "vox" to the lower
resolution one. In your case of course these should be the same as you
only have the 0.7 mm one.
The structure matlab variable is called "transform". I have put the
description of the fields in form of code , so potentially you could run
it and create the structure for the transform file.
Its fields are:
*%==============================================================*
*transform.vox07mm_filename *= 'T1w_acpc_dc_restore.nii.gz'; % This is
the name of the HCP processed stuctural MRI with 0.7mm resolution
*transform.vox_filename* = transform.vox07mm_filename ; % This is
supposed to be the lower resolution version used for MEG but here you
assign it also to the 0.7mm
*transform.vox07mm2spm* = vox07mm2spm; %This is the transformation
matrix from voxel to "spm" coordinates mentioned earlier retrieved from
the 0.7mm structural MRI.
*transform.spm2vox07mm* = inv(transform.vox2spm); % This is the inverse
transformation of the above from "spm" coordinates to voxels.
*transform.vox2spm *=transform.vox07mm2spm; %This is the transformation
matrix from voxel to "spm" coordinates mentioned earlier . As you only
have the 0.7mm structural here is the same also for the lower resolution
case.
*transform.spm2vox* = inv(transform.vox2spm); % This is the inverse
transformation of the above from "spm" coordinates to voxels.
*transform.spm2bti* = spm2bti; % This is the transformation between spm
and bti coordinate systems. In this case it is assumed they are the same
so it the identity matrix created earlier.
*transform.bti2spm* = bti2spm; %Same case here
*transform.vox07mm2bti* = transform.vox07mm2spm ; % As spm and bti are
identical coordinate system this is just the transformation from voxels
to "spm" coordinates already created;
*transform.bti2vox07mm *= transform.spm2vox07mm; % The same is true for
the inverse transformation.
*transform.vox2bti* = transform.vox2spm; % Same for the lower
resolution case used for MEG, which of course yo dont have, as yo uonly
have the 0.7 mm so this is the same as for the 0.7mm case
*transform.bti2vox *= transform.spm2vox; % Same for the inverse
transform of the lower resolution case
*transform.vox2bti_registered* = transform.vox2bti; % This is an
intermediate transformation from voxel to bti used only for MEG purposed
but for your case should be exactly the same as the voxel to bti
transformation which of course is the same as the voxel to spm
transformation as these coordinate systems are the same in your case.
*%==============================================================*
Once you have created the "transform" variable you have to write it in
an ascii file, which is not as strightforward as writing a numberical
array in an ascii in matlab.
The easiest way is to download the MEG pipeline code from ConnectomeDB
and use the function from the "analysis_functions" subfolder called
"hcp_write_ascii.m"
simply as
transformfile=<subj>_MEG_anatomy_transform.txt
hcp_write_ascii (transformfile,'transform');
and
I believe that should do it.
Give it a try and let me know if it worked or not .
It might be of interest to other users too.
Have a nice day
Best
Giorgos
On 2/10/2017 10:41 PM, K Jeffrey Eriksen wrote:
Hi,
I am trying to import HCP processed data into Brainstorm, a
Matlab-based package for EEG/MEG source analysis, in order to build an
EEG source model. I do not work with MEG.
Unfortunately I am failing since I do not have the
<subj>_MEG_anatomy_transform.txt file that is created during the
HCP-MEG pipeline processing. Can someone help me figure out a way to
create this transform for non-MEG HCP processed subjects? In our lab
we use the WUSTL HCP pipeline with some local modifications.
Thanks,
-Jeff
_______________________________________________
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
