Dear All,

I have a quick about dimension deduction. I now have two images. The first one is subject level smoothed data, and the other one is group level ROI-to-voxel connectivity mask. The mask was exported and binarized in order to apply this binarized mask on subject level smoothed data for the purpose of power calculation. Now my questions is, when I used matlab to import the data, the subject level smoothed data has [91,109,91] data points with the volume variable, then the group level mask has [74,92,78] data points. It suggested to me that some margin slides were cut of in the process of noising/Level1/Level2. I want to know how the margin slides were cut? For example, something like 10 top slides, 9 bottom slides, 5 left, and 5 right?

Also, I'm making matlab code to do power calculation. I want to get the information on, for example, in one ROI, the power distribution across a certain band. If anyone knows a better way to do this, suggestions are very welcomed too!

Thanks a lot!

Haoyun

Dear Haoyun,

That information is stored as part of the header information in the corresponding files, in particular the mat field in the structure returned by spm_vol represents a 4x4 affine transformation between individual voxels in those 3d volumes and typically-common x/y/z world coordinates (e.g. MNI-space), so, for example, the following code allows you to find the correspondence between voxel coordinates in one file and the other:

vol1 = spm_vol( file1 );
vol2 = spm_vol( file2 );
voxel_coords1 = [i; j; k; 1];
voxel_coords2 = inv(vol2.mat) * vol1.mat * voxel_coords1;

In any way, perhaps a simpler way to extract the subject-level smoothed BOLD data from your ROI mask would be to use the syntax:

x = rex( fileData, fileMask);

which will extract the average BOLD signal timeseries stored in file "fileData" within the region of interest defined in your "fileMask" file (and that will take into account the above header information to find the appropriate match between the two files)

Hope this helps
Alfonso
Originally posted by Sophie Zhang: