Scoring phenotypes based on two fluorescence channels

I am using a system, where I stain for ABCG2 in one channel and DAPI in another channel. Because of the granular nature of ABCG2, it is impossible to use this channel alone to count positive cells. I think that if the DAPI channel could somehow be superimposed on the ABCG2 channel, it would be easier to detect number of cells expressing ABCG2? Can this be done in CellProfiler Analyst, or do any of you guys know alternatives to count this in an automated fashion?

I already know how to use CPA for machine-learning in a single channel…



Hi Chris,

I’m terribly sorry for the slow reply, but I’ve been on vacation for two weeks. I’m not sure what you’re asking in your post though… are you asking if you can simply superimpose the channels of the image in CPA? (If so the answer is yes.)

However, it sounds like you’re asking how to use the actual data in these channels for training… if that’s your question, then the short answer is that you’d use CellProfiler to measure the objects in each of your channels and then you write this data to the database where CPA’s machine learning algorithm can chew on it.


I have a question that seems to go a long with this. I am having trouble with the cell classifier crop images that come up in the training for phenotypes. I am trying to classify cell phenotypes but the cropped images are too cropped so it only gives me the nuclei. the images are taken at 10x and the cells are GFP tagged and DAPI stained. I am hoping to classify cells that are in the process of dividing. Any suggestion would be grateful!

It sounds like you just need to increase the value of the image_tile_size property in your properties file.