Dear List,
I have a client who is interested in segmenting some brightfield images of
Trichrome stained muscle cross-sections in order to set up some automated
cell counting. Unfortunately, the test images that he has given me to work
with have been refractory to my initial attempts at histogram or color based
segmentation. I've also tried applying color channel separation, some image
arithmetic and a variety of edge detection filters and tried a few
segmentation plugins that I downloaded from the image J website (incl RATS,
k-means clustering) to no avail. The images look very similar to
http://moon.ouhsc.edu/kfung/jty1/Com04/Com04Image/Com411-1-05.gif Does
anyone have a segmentation procedure or plugin that can properly segment out
these cells to generate a count that is reasonably accurate wrt what one can
make out pretty readily by eye? It seems to me that it should be possible.
I can provide a representative TIFF file to anyone who is interested in
following up with me directly.
Thank you for your attention,
Aleksandrs J. Spurmanis
Microscopy Specialist
Imaging Facility
McGill University Life Sciences Complex
Francesco Bellini Building
3649 Sir William Osler
Suite 137
Montreal, QC
H3G 0B1
tel.: (514)-398-5248
fax: (514)-398-7452
[hidden email]
http://www.mcgill.ca/lifesciencescomplex/core/imaging/
Locked
