Generating image alignment

cellprofiler

#1

I am using CellProfiler to automatically generate alignments from the same tissue. I am trying to match and align the first hematoxylin stain to an IHC image. However, with the pipeline that I am currently using, sometimes the alignments are inaccurate (see image)

How can I improve my pipeline and make the automatic alignment generations more accurate? When I am going to further analyze these images, the images won’t match and I will be unable to get a good analysis. Is there any other way I can do this using a different software (i.e. Fiji)?


#2

I would investigate doing it the way round: IHC + haematoxylin stain in the same tissue sample, then use colour deconvolution to separate the dyes. You can do that in ImageJ and Fiji (and perhaps in cellprofiler, but I do not know).
http://www.mecourse.com/landinig/software/cdeconv/cdeconv.html


#3

Hi,

Can you post a pipeline and the image set? Otherwise, I’d make sure to use the slower (but more accurate) 'Mutual Information" version of alignment. That being said, your images are very different, so it may fail sometimes.

You can indeed do color deconvolution of your multiply-stained tissues in CP with the “UnmixColors” module.