How to make ROI fit for function to avoid error with subscript indices?

I keep getting the following error when trying to use another ROI than the one that was made when I wrote the function and script:

Subscript indices must either be real positive integers or logicals.

Error in getROIvals (line 10)
   roi_vals(i) = mean(cur_img(ROI));

Error in roi_fit (line 12)
roi_vals = getROIvals(data,slice,ROI);

Can somebody please point out how to fix it? I originally used a ROI = roipoly(mat2gray(S0)) which has size 128x128, so I thought there would be no problem using another ROI-mask with same size and format (perhaps the problem is that it's not same size and format, allthough it is 128x128).

The function is:

function roi_vals = getROIvals(data,slice,ROI) 
S = double(squeeze(data(:,:,slice,:))); %size 128x128x8

n = size(S,3);

roi_vals = zeros(n,1);

for i = 1:n
   cur_img = S(:,:,i);
   roi_vals(i) = mean(cur_img(ROI));   
end

end

What does not work:

slice = 3; %the slice the mask is applied to 
data = load4D('VOLUME_IMAGEb0',8,[0 50 100 150 400 500 600 800]);
%S0 = double(squeeze(data(:,:,slice,1))); % for the ROI = roipoly(mat2gray(S0)), which works..

ROI = mat2gray(BW); %BW is a 128x128 mask yielded from another script
imagesc(ROI);
save('ROI');

roi_vals = getROIvals(data,slice,ROI);

Answer 1

[BME student, Copenhagen]

Actually, I looked at ROI created with the roipoly the first time, and it created 128x128 logical, so I solved it by setting ROI = logical(mat2gray(BW)) and now the little "blip" is fixed, so the whole machinery works and I can get som serious data output! Yay!