Hi,

Yes, the xyz variable contains the centroid of each of the ROIs.

If the atlas file is 4D, you can try SPM's 4D to 3D utility to quickly get to separate ROI files. In either case, this can be achieved using something on the lines of:

%% This case is for 3D atlas file having unique intensity values for each ROI
% Read in atlas.nii file
vol_atlas   = spm_vol('atlas.nii');
data_atlas = spm_read_vols(vol_atlas);
data_tmp  = data_atlas;

% Get all unique intensities
roi_vals    = unique(data_atlas);
num_vals = length(roi_vals);

% Loop over each intensity
for i = 1:num_vals

% Binarize the values
data_tmp(data_tmp~=roi_vals(i)) = 0;
data_tmp(data_tmp==roi_vals(i)) = 1;

% Write out
roi_name = ['ROI_', num2str(i, '%03d'), '.nii'];
vol_atlas.fname = roi_name;
spm_write_vol(vol_atlas, data_tmp);

% Reassign original values to data_tmp
data_tmp = data_atlas;
end

% ----------------------------------------------------------------------

%% This is the case of 4D atlas file where each volume is an ROI
% Read in atlas.nii file
vol_atlas = spm_vol('atlas.nii');
data_atlas = spm_read_vols(vol_atlas);

% Get number of ROIs
num_rois = size(data_atlas,4);

% Loop over each volume and write it out
for i = 1:num_rois
roi_name  = ['ROI_', num2str(i, '%03d'), '.nii'];
vol_atlas(i).fname = roi_name;

% If a field n exists, set it to 1
if isfield(vol_atlas, 'n')
vol_atlas(i).n = [1,1];
end

spm_write_vol(vol_atlas(i), squeeze(data_atlas(:,:,:,i)));
end


As an additional note, it is worthwhile to change the data type (vol.dt) as all ROIs from above are binary. Also the .info file contains a lot of useful information. The dimension of the overall file can be picked up from the NIfTI header.


Hope this helps


Best
Pravesh


Originally posted by Hyunwoo Kim: