Welcome to CellProfiler! This example is similar to your issue, but in any case, here’s how I would do it for your data (see attached pipeline).
The basic idea is to segment nuclei, then “grow” them out blindly into “cytoplasm” and then make measurements in the green channel on the nuclear and cytoplasm compartments. We have to grow out blindly since some cells don’t have marker in their cytoplasm.
(1) Images: Load your images. Side note: avoid JPG as they are lossy. Try PNG or TIFF.
(2) NamesAndTypes: Be sure to load your color images as color here. Conversely, if you have grayscale images which CellProfiler needs anyway, change the setting here.
(3) ColorToGray: Splits the image colors into grayscale. Side note: your red image might be expected to have nothing in it, but there is some info there. Better to image directly into grayscale if possible.
(4) IdentifyPrimaryObjects: (Try and) segment the nuclei. Your nuclear marker unfortunately has a fair amount of variability and it bleeds into the cytoplasm, so this makes it challenging to get all the nuclei while also not segmenting the cytoplasm. I did the best I could without more time investment.
(5) IdentifySecondaryObjects: Grow out (blindly) 50 pixels. Adjust the size as you see fit, which should approximate the cyto width of your cells.
(6) IDTertiary: this defines cytoplasm, i.e. without nuclei.
(8) DisplayDataOnImage: simply there to visualize decent thresholds for the ClassifyObjects module
(9) ClassifyObjects: Create bins to classify objects based on the intensity metrics that you choose.
Hope that helps!
DL_filtering_by_compartment.cppipe (12.6 KB)