within_voxel_z_scoring.m

function within_voxel_z_scoring(list_images, mask_path)
% Perform within-voxel z-scoring throughout an entire list of images
% FORMAT within_voxel_z_scoring(list_images, mask)
% list_images      - cells column, each cell contains the path of one image
% mask_path        - path to a binary mask to constrain the area where 
%                   z-scoring is performed. It should be the same mask used
%                   for the univariate model used as input to the
%                   multivariate analysis
%                   
%
%__________________________________________________________________________
%
% The function does not return any output but each image that has been z-scored
% is saved in the exact same folder as the original with the exact same 
% name but starting with the prefix 'z_'
hmask = spm_vol(mask_path);
[mask, cmask] = spm_read_vols(hmask);
cmask(4,:) = 1;
voxmask = hmask.mat\cmask;
voxmask = voxmask(:,mask > .1);

im_mean = nan(size(mask));
im_std = nan(size(mask));
mattmp = spm_get_data(list_images,voxmask(1:3,:));
mattmp_mean = mean(mattmp,1);
mattmp_sd = std(mattmp,1);

im_mean(mask > .1) = mattmp_mean;
im_std(mask > .1) = mattmp_sd;

    for j = 1:size(list_images,1)
        htmp = spm_vol(list_images{j});
        voltmp = spm_read_vols(htmp, mask);
        voltmp = (voltmp - im_mean)./im_std;
        [htmppathstr, htmpname, htmpext] = fileparts(htmp.fname);
        htmp.fname = fullfile(htmppathstr, ['z_' htmpname htmpext]);
        htmp.dt = [64 0];
        spm_write_vol(htmp, voltmp);
        fprintf('saving %s \n', htmp.fname)
    end
end