find mean of 2d slice in 3d segmented volume

I have a 3d variable with slices containing segmented rois (128x128x10) . I am trying to find the mean value for each slice in such a way that it does not average the zero pixels around the roi. i want a 1 x 10 array containing the mean value for each slice.

[Turlough Hughes]

How did you get your roi?

[Raheema Al Karim Damani]

I used roi poly. and then I masked the original image onto the binary segment

Answer 1

[Image Analyst]

Try this:

[rows, columns, slices] = size(image3d);
% image3d is your masked gray level image with some zeros in it.
sliceMeans = zeros(1, slices);
for slice = 1 : slices
    % Get this slice
    thisSlice = image3d(:, :, slice);
    % Find out what pixels are nonzero
    nonZeroIndexes = thisSlice ~= 0; 
    % Compute the means of only non-zero pixels.
    sliceMeans(slice) = mean(thisSlice(nonZeroIndexes))
end

[Raheema Al Karim Damani]

thank you! it works

Answer 2

[Turlough Hughes]

Ok I think I know what you want. Looking at the documentation for roipoly, I assume you got a binary image representing the region of interest, lets call that BW. Let's also say your images are in the variable, V, which is 128 by 128 by 10. To determine the mean value of each slice for the ROI only, you could do the following:

[indRows,indCols] = find(BW); % where BW is the binary image representing your roi.
result = mean(V(indRows,indCols,:),[1 2]);

So the idea here is to find row and column indices for your ROI (I'm also assuming it is the same roi for each slice). Once you know your indices you its simply a case of plugging the values into the mean function.

[Turlough Hughes]

You may also want to reshape the result;

result = reshape(result,1,10)

[Raheema Al Karim Damani]

Thank you for your response! It is actually diff rois for each slice, hence diff dimensions.