Skip to content

Latest commit

 

History

History

create_library

README.md

Create an MRI library for MEMES

MRI Estimation for MEG Sourcespace (MEMES) is a set of tools for estimating an appropriate structural MRI for MEG source analysis in Fieldtrip and/or SPM.

Please follow these instructions to create a compatible MRI database for MEMES. You will need the Fieldtrip toolbox in your MATLAB path.

MEMES

Database Organisation

Each MRI database should be organised as below, with each subject in separately named folder, no spaces. It would probably help to have some sort of consistent naming practice, but MEMES doesn't really care too much.

Try to keep your MRI database folder tidy - no stray files or folders!

Org1

Each subject folder should contain the following:

Org2

(the SLIM-T1w.nii is the original MRI for this subject)

Initial Realign

In order to improve the accuracy of MEMES it is important to realign by marking the Nasion, LPA and RPA.

Mesh Creation

The realigned MRI can now be used to create a 3D cortical mesh of the subject's scalp. This is then matched with Polhemus data by MEMES. Steps are below:

%% Segment the MRI leaving the scalp
cfg = [];
cfg.output    = 'scalp';
cfg.scalpsmooth = 5;
cfg.scalpthreshold = 0.08;
scalp  = ft_volumesegment(cfg, mri_realigned);
    
%% Create mesh out of scalp surface
cfg = [];
cfg.method = 'isosurface';
cfg.numvertices = 10000;
mesh = ft_prepare_mesh(cfg,scalp);
mesh = ft_convert_units(mesh,'mm');

Note 1: You may need to alter the cfg.scalpthreshold = 0.08 parameter for some subjects. Always plot this for quality management and adjust the threshold as needed.

Note 2: You can make the mesh more or less dense by altering the cfg.numvertices = 10000 parameter. However 10,000 seems to work well for MEMES as a balance between and accuracy computational efficiency.

Headmodel and Sourcemodel

The realigned MRI can now be used to create a singleshell headmodel:

%% Segment the brain
cfg           = [];
cfg.output    = 'brain';
mri_segmented  = ft_volumesegment(cfg, mri_realigned);
    
%% Create singleshell headmodel
cfg = [];
cfg.tissue = 'brain';
cfg.method='singleshell';
headmodel = ft_prepare_headmodel(cfg, mri_segmented);

% Save headmodel
save headmodel headmodel;

5,8 and 10mm sourcemodels aligned to MNI space also need to be crea:

%% Create sourcemodel
    
sourcemodel_mm = [10 8 5];
    
for size = 1:length(sourcemodel_mm);
        
    fprintf('Creating %dmm sourcemodel\n');
        
    % create the subject specific grid, using the template grid that has just been created
    cfg                = [];
    cfg.grid.warpmni   = 'yes';
    cfg.grid.resolution = sourcemodel_mm(size);
    cfg.grid.nonlinear = 'yes'; % use non-linear normalization
    cfg.mri            = mri_realigned;
    cfg.grid.unit      ='mm';
    cfg.inwardshift = '-1.5';
    grid               = ft_prepare_sourcemodel(cfg);
        
    % Save
    sourcemodel3d = grid;
    save(sprintf('sourcemodel3d_%dmm',sourcemodel_mm(size)),'sourcemodel3d');
        
    % Clear for next loop
    clear grid sourcemodel3d
end