Georgos,
Understood about the naming.
Next question: I have found the Brainstorm code that reads the
transform file:
%% ===== MRI=>MNI TRANSFORMATION =====
% Convert transformations from "Brainstorm MRI" to "FieldTrip voxel"
Tbst2ft = [diag([-1, 1, 1] ./ sMri.Voxsize), [size(sMri.Cube,1); 0;
0]; 0 0 0 1];
% Set the MNI=>SCS transformation in the MRI
Tmni = transform.vox07mm2spm * Tbst2ft;
sMri.NCS.R = Tmni(1:3,1:3);
sMri.NCS.T = Tmni(1:3,4);
% MNI coordinates for the AC/PC/IH fiducials
AC = [0, 3, -4] ./ 1000;
PC = [0, -25, -2] ./ 1000;
IH = [0, -10, 60] ./ 1000;
Origin = [0, 0, 0];
% Convert: MNI (meters) => MRI (millimeters)
sMri.NCS.AC = cs_convert(sMri, 'mni', 'mri', AC) .* 1000;
sMri.NCS.PC = cs_convert(sMri, 'mni', 'mri', PC) .* 1000;
sMri.NCS.IH = cs_convert(sMri, 'mni', 'mri', IH) .* 1000;
sMri.NCS.Origin = cs_convert(sMri, 'mni', 'mri', Origin) .* 1000;
%% ===== MRI=>SCS TRANSFORMATION =====
% Set the MRI=>SCS transformation in the MRI
Tscs = transform.vox07mm2bti * Tbst2ft;
sMri.SCS.R = Tscs(1:3,1:3);
sMri.SCS.T = Tscs(1:3,4);
% Standard positions for the SCS fiducials
NAS = [90, 0, 0] ./ 1000;
LPA = [ 0, 75, 0] ./ 1000;
RPA = [ 0, -75, 0] ./ 1000;
Origin = [0, 0, 0];
% Convert: SCS (meters) => MRI (millimeters)
sMri.SCS.NAS = cs_convert(sMri, 'scs', 'mri', NAS) .* 1000;
sMri.SCS.LPA = cs_convert(sMri, 'scs', 'mri', LPA) .* 1000;
sMri.SCS.RPA = cs_convert(sMri, 'scs', 'mri', RPA) .* 1000;
sMri.SCS.Origin = cs_convert(sMri, 'scs', 'mri', Origin) .* 1000;
% Save MRI structure (with fiducials)
bst_save(BstMriFile, sMri, 'v7');
and here is the relevant contents of the transform file I got with the
HCP-MEG example recommended by BS:
transform.vox07mm_filename = 'T1w_acpc_dc_restore.nii.gz';
transform.vox_filename = '175237_MEG_anatomy_anatomical.nii';
transform.vox07mm2spm = [
-0.7 0 0 90.7
0 0.7 0 -126.7
0 0 0.7 -72.7
0 0 0 1
];
transform.vox07mm2bti = [
0.0242485 0.689697 -0.143363 -99.3102
0.702637 -0.0220904 0.00540799 -88.0016
-5.54434e-05 0.143686 0.684532 -77.7522
0 0 0 1
];
So only two of the transforms in the transform.vox_filename =
'<subjID>_MEG_anatomy_anatomical.txt' are needed by BS. I just have to
figure out how to create them for my non-MEG data. One disconnect I am
having is that below (from your earlier email) you say these two are
identical, but BS obviously does not think so as they are obviously
different in the above snippet:
>>transform.vox07mm2spm = vox07mm2spm; %This is the transformation
matrix from voxel to "spm" coordinates mentioned earlier retrieved
from the 0.7mm structural MRI.
>>transform.vox07mm2bti = transform.vox07mm2spm ; % As spm and bti are
identical coordinate system this is just the transformation from
voxels to "spm" coordinates already created;
I do not process MRI data myself, concentrating on the EEG analysis)
so do not know how to setup and run the HCP pipeline or the special
HCP_MEG pipeline. I did try searching all the HCP_MEG pipeline M-files
and found no occurrence of the transforms above, so they must be
embedded in the shell scripts or compiled C code somewhere, presumably.
Can you comment on the above and perhaps point me in the right
direction? One idea I have is to try to run the HCP-MEG pipeline minus
the actual MEG portions. Another is to find out if the standard HCP
pipeline produces the transforms I need.
Thanks,
-Jeff
*From:*Michalareas, Giorgos [mailto:giorgos.michalar...@esi-frankfurt.de]
*Sent:* Friday, February 10, 2017 4:50 PM
*To:* K Jeffrey Eriksen
*Subject:* RE: [HCP-Users] MEG_anatomy_transform.txt
I think this should not be a problem.
Just use the T1 you have but make sure the fields names of the
"transform" structure remain the same as in the description I sent you
because I am pretty confident that Brainstorm is expecting to find
these exact filenames,
Give it a try snd let me know how it goes
Best
Giorgos
------------------------------------------------------------------------
*From:*K Jeffrey Eriksen [eriks...@ohsu.edu]
*Sent:* Saturday, February 11, 2017 1:02 AM
*To:* Michalareas, Giorgos
*Subject:* RE: [HCP-Users] MEG_anatomy_transform.txt
Georgios,
I have read through this now, and have one follow-up question. While
we are using the HCP pipeline, our original T1s are 1.0 mm cubic
voxels. Could you please comment on this, or possibly modify what you
have written to accommodate this fact?
Thanks,
-Jeff
*From:*hcp-users-boun...@humanconnectome.org
<mailto:hcp-users-boun...@humanconnectome.org>
[mailto:hcp-users-boun...@humanconnectome.org] *On Behalf Of *Georgios
Michalareas
*Sent:* Friday, February 10, 2017 3:05 PM
*To:* hcp-users@humanconnectome.org <mailto:hcp-users@humanconnectome.org>
*Subject:* Re: [HCP-Users] MEG_anatomy_transform.txt
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 <mailto:HCP-Users@humanconnectome.org>
http://lists.humanconnectome.org/mailman/listinfo/hcp-users
_______________________________________________
HCP-Users mailing list
HCP-Users@humanconnectome.org <mailto:HCP-Users@humanconnectome.org>
http://lists.humanconnectome.org/mailman/listinfo/hcp-users
