I am currently working on a project involving 3D image preprocessing and nuclear segmentation, and I would appreciate any guidance or suggestions you might have on preprocessing and segmentation for nucleus on tissue sample.
Specifically, my workflow involves:
- Identifying maxima,
- Analyzing particles to measure intensity,
- Calculating the foci number
Any help or clarification regarding this workflow would be invaluable to me.
You mention 3D so I assume you have a stack of slices. How is the resolution in the 3rd dimension?
I would try, in 3D:
- Throw away the red and green channels; no info there.
- Find a good threshold so that all or "enough" bright spots in the nuclei are above the threshold. Never mind that you also get small bright spots outside of the nuclei as long as those are much smaller than the spots inside the nuclei.
- Count the number of voxels of each spot and find a good threshold so that you can discard all the small spots.
- Do an Euclidean distance transform. The value of all voxels will then be set to their distance to a bright nuclear spot.
- Do marching cubes at some small distance. Large enough so that the resulting surfaces do not "creep inside" the nuclei, but small enough so that nuclei will not be joined.
Note that the surfaces may appear too much like raspberries.
I don't know whether Fiji/ImageJ is able to do it. Octave (or MatLab) certainly can.
Good luck!
- Badges
- Science topic
- Similar topics
- Biological Science
- Histology