Hi all,
I am a very new user of ImageJ. Currently I am trying to use the skeleton analysis to analyse cell morphology under different experimental conditions. But I encounter a few problems and I shall be very glad if I can get some suggestions from you.
I follow from the procedure of a paper which is briefly:1) Converting the specifically-stained (DAB) light-microscopy RPG image to 8-bit.2) Setting a threshold to generate a binary image.3) Skeletonize.4) Analyze Skeleton.5) Manually count the number of cells6) Compare number of process ends/cell, length of process/cell, number of junctions/cell,etc
It is pretty straightward, except that I can't find a suitable threshold. Under some conditions, the cells are denser, bigger and more intensely stained; under others the cells are sparse and weakly stained. So if I am to keep a constant threshold across my analysis, some images will have lots of noise (and background signals that join together), causing unreliable counts on branch length, and process ends. On the other hand, some images with weakly stained or sparse cells may have the cell processes broken up due to the high threshold, causing inaccuracy in parameters such as process ends again.
So I am wondering if there is a solution for this? I have also tried to use particle analyzer to pick away the 1 pixel noises before converting to skeleton. But then where shall I set the cut-off as noise?
Another question is, is there any way for Image J to recognize or count cell for me? I am working on microglia, so they have lots of branches, and may appear as joining together at the 40X magnification that I am currently using.
Thank you very much for the help.
Cheers,Tom Cheng
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
