Colocalization analysis for apoptosis assay

Hi there. I am trying to analyze these two images for colocalization. I am trying to determine alive (green) vs necrotic (red) vs apoptotic (yellow) I have seen many papers that seem to use ImageJ/Fiji to analyze AO EB staining but cannot seem to find a suitable method. I have tried Ezcolocalization JACoP and coloc2. Any advice would be greatly appreciated. Thanks!
3.TIF (14.1 MB) 4.TIF (14.1 MB)

It generally helps to have a blue channel for experiments like this to determine where the nuclei are. Then you can ask two questions, where are my blue cell objects, then what color are my blue objects.

Without the base channel, you have some red objects that are red, some that are red-green, and some that are green, and no single method is going to pick them all up (though you can try pixel classifiers or adding the two channels together).

I would also recommend converting your images into 8bit mono images, what you have posted are two RGB images with no pixel size metadata.

I used QuPath for this instead, but you could probably follow a similar train of steps using something like Weka pixel classifier to find your initial cell ROIs.

Pixel classifier to find nuclei (poorly, I did not take much time):

Very, very not optimized. But also easier with Hoechst/DAPI depending on whether cells are live or fixed.

Convert those masks into ROIs.

Add measurements to the objects for each channel.

Then classify all of the objects as single positive or double positive per channel based on the mean or median intensity.

If you want to do this in QuPath I can point to the readthedocs, but I would use a similar workflow for FIJI. I don’t think you actually need to measure co localization (based on your description) so much as classify cells as positive or negative in each channel per marker.

Though again, determining the boundaries would be easier with a blue channel :slight_smile:

Thanks for the feedback. There isn’t specifically a blue channel to identify cells. If the cells are necrotic they stain red, if they are alive they stain green and various stages of apoptosis stain somewhere between. I could use the composite image as to identify all cells and create ROIs then use this to compare against each image. With all the plugins I’ve tried I always background subtract and covert to 8 bit. I just uploaded the tifs as they are collected from the microscope. The papers I have looked at make the method seem quite simple. I just cant seem to figure it out.