How to do segmentation of the overlapped cells in an image and obtain their co-ordinate?
2 views (last 30 days)
Show older comments
I have two images one for cell and other for corresponding nuclei. My criteria for overlapping is that if the any cell is shown as a single connected component but their are more than one nuclie enclosed in the boundary of the connected component of cell. Suppose, If I have two nuclie in a connected component of cell then we have to separate that cell region into two connected components each having one nuclie and similarly for if more nuclie. I have tried used watershed algorithim in the IPT but didn't make much out of it. Please help me out to segment only such overlapping cells to their corresponding nuclei and not the ones in there is only one nuclei for one cell. (RUN THE CODE ONCE) Thanks in advance.
<<
>>
nucleus_gray = imread('nuclei.jpg'); cell_gray = imread('cell.jpg');
level = 0.17; Ithresh_nucleus_1 = im2bw(Inucleusgray, level);
level2 = 0.18; Ithresh_cell_1 = im2bw(Icellgray, level2); Ifilled = imfill(Ithresh_cell_1, 'holes'); Ithresh_cell_1 = bwareaopen(Ifilled, 245);
s = regionprops(Ithresh_nucleus_1, 'centroid'); centroids = cat(1, s.Centroid);
imshow(Ithresh_cell_1) hold on plot(centroids(:,1), centroids(:,2), 'b*') hold off
3 Comments
Answers (1)
Siyu Guo
on 29 Apr 2018
I can't see how you have done the watershed. It seems to me that an Euclidean distance transformation on the binary nuclei image should be a proper landscape for watershed: d = bwdist(BW_nuclei); L = watershed(d); And afterwards, compute the logic AND of each watershed region with the segmented binary cell image. I don't have the necessary images, so I can just provide a possible way for you to try.
See Also
Products
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!