Why are these codes NOT giving 4 biggest connected component in the stack image ???

Hi,

I need your help. I have a tiff image that includes 7 slices. I want to find 4 biggest connected component in the tiff image. These components might be in one or two slices; Or each of these components might be in different slices. Therefore, I used bwlabeln to label components in the tiff image and wrote the following codes to get the first biggest connected component (If I find the first connected component then I apply the same codes to find the 2nd, 3th and 4th biggest connected component). But it does not give correct result

Codes are,

 [temp77 NumberOfConnComps] = bwlabeln(imgStack); 

 for j=1:NumberOfConnComps
               temp88 = temp77==j;
               part(j)=sum(temp88(:));

      end

big=max(part);
k2=find(part==big);
biggestPart=temp77==k2;

However, the above codes do NOT give the first biggest connected component. When I write the following codes to see the 1st biggest connected component, then I see that it includes other components :(

 for slice=1:7
  if (slice == 1)
     imwrite(squeeze(biggestPart(:,:,slice)),['biggestPart.tif'],'tif', 'WriteMode', 'overwrite','Compression', 'none');
  else
     imwrite(squeeze(biggestPart(:,:,slice)),['biggestPart.tif'],'tif', 'WriteMode', 'append','Compression', 'none'); 
  end

end

WHY is it giving false result ? How to solve this problem ?

Urgent help please

Thanks

Answer 1

[S C.Carl]

I added these codes, but the result does not changed;

 for slice=1:7
   if temp77(:,:,slice)==k2 
          biggestPart = temp77(:,:,slice)==k2;

     if (slice == 1)
        imwrite(squeeze(biggestPart(:,:,slice)),['biggestPart.tif'],'tif', 'WriteMode', 'overwrite','Compression', 'none');
     else
         imwrite(squeeze(biggestPart(:,:,slice)),['biggestPart.tif'],'tif', 'WriteMode', 'append','Compression', 'none'); 
     end
   end
end

When I look at 2D slices separately, I can see components (there are 49 connected components). However, when I run the above codes and open the saved "biggestPart.tif" image, then I saw small connected components. The above codes should give only one (the first biggest connected) component. Right ? What is the wrong in the above codes ?

Answer 2

[Image Analyst]

What you want to do is use bwareafilt(). You can get the largest blob this way

bw1 = bwareafilt(imgStack, 1, 'largest');

Then write it out or do whatever you're going to do with it. Then to get the second largest blob, etc., all by itself in a 3D image, you'd do

bw2 = bwareafilt(imgStack - BW1, 1, 'largest'); % Get second largest blob.
bw3 = bwareafilt(imgStack - BW2, 1, 'largest'); % Get third largest blob.
bw4 = bwareafilt(imgStack - BW3, 1, 'largest'); % Get fourth largest blob.

Another way is to get all the areas, sort them, and then extract them in order with ismember():

% Measure all the volumes.
measurements = regionprops(temp77, 'Area');
% Get all areas
allAreas = [measurements.Area];
% Sort them in descending order
[sortedVolumes, sortIndices] = sort(allAreas, 'Descend');
% Now relabel the labeled image with intlut
labeledImage = intlut(temp77, sortIndices);
% Now get largest blobs
big1 = ismember(labeledImage, 1);
big2 = ismember(labeledImage, 2);
big3 = ismember(labeledImage, 3);
big4 = ismember(labeledImage, 4);

Not sure which way is faster. I have not tested either - it's just off the top of my head so if you run into problems, ask.

[Image Analyst]

S.C.Carl's "Answer" moved here:

Dear Image Analyst Thank you very much for this response. However, I dont have bwareafilt() function :( I couldnot find it in mathworks.com :(

What can I do ? :(((

[Image Analyst]

I don't see bwareafilt() in my second suggestion. Did you see the part of my answer after I say "another way...."

For a 2-D example, see my attached demo.

Of course the best way is to just upgrade to the latest version.

[S C.Carl]

Dear ImageAnalyst,

Thanks for your support, I solved the problem with your help. I will upgrade the matlab version as soon as possible Thanks again Best wishes

[Image Analyst]

If the problem is now solved, and I helped with that, then could you "Accept this answer" to give me credit? Thanks in advance.