Quantification of Co-localized TH and DAPI in Confocal Stacks

I am a new user (and, therefore, cannot upload attachments). The goal is to use Cell Profiler to quantify the percentage of the total cells (total DAPI) that are TH positive. I have z-stacks with a TH channel and a DAPI channel. The basic Pipeline is: Threshold for TH to eliminate faint TH staining in the neurites, Identify Primary Objects for DAPI, Identify Primary Objects for TH, Relate Objects (TH Parent - DAPI Child), Classify Objects. I am experiencing a few issues: (1) I am having trouble accurately segmenting TH cell bodies. Thresholding eliminates the TH staining in the neurites but it also eliminates more faint TH cell bodies. I tried setting it up such that the TH cell bodies are Secondary Objects to the DAPI Primary Objects but that was not effective. (2) I am having trouble with the Relate and Classify Object Modules; specifically, they are not yielding the output that I would like; namely, count for TH-DAPI doubles and DAPI singles. Any tips or suggestions are appreciated.



Hard to diagnose without images or a pipeline, but some general guidance-

  1. How many different thresholding methods have you tried? I’d carefully read the different settings for Threshold to get a sense of how the different methods perform in different ways, as well as the different settings (such as the Threshold Correction Factor) that you can use for making tweaks.

  2. If your TH signal is present in the nucleus (which I’m guessing it is since you say it’s co-localized), a much simpler pipeline would be to do Threshold->IdentifyPrimary for DAPI->MaskObjects using the ‘Mask based on Image’, and ‘Remove depending on overlap’ options (read the help!) to keep only nuclei that overlap with the thresholded TH signal by a certain amount (20%, 60%, whatever seems to produce the most accurate results). Since you can set the AMOUNT of overlap that triggers being classified as positive by doing it this way, you can probably afford to set your threshold for TH-positive areas in 1) lower since a stray neurite crossing a nucleus won’t be enough to classify the nucleus as positive.

Once you’ve done that, either your ExportToDatabase ‘Image’ table or your ExportToSpreadsheet ‘Image.csv’ spreadsheet should have columns for Count_Nuclei and Count_ThPositiveNuclei (or whatever you call your classes).

1 Like

Thanks very much for your response. I’ve modified the Pipeline (see link below) as per your (very good) suggestions. The TH signal is not in the nucleus. It can be found throughout the cytoplasm and fills cell bodies and neurites. When dopamine neurons are stained for TH and any nuclear marker, often what you will see is a TH negative nucleus with TH throughout the cell body and neurites.

I’m concerned about Thresholding as it removes the TH signal in the neurites (which is desirable), but also removes faint TH positive cell bodies (which is not desirable). Is there any way to remove the TH positive neurites while sparing the faint TH cell bodies?

Pipeline and 2D Images

If it’s found in the cytoplasm, can you use ExpandOrShrinkObjects to expand the nuclei by a few pixels then test for % positive thresholding?

1 Like

Hi. I added ExpandOrShrinkObjects to expand the nuclei as per your suggestion then tested the modified Pipeline on a handful of image sets. The modified Pipeline seems to work well with regard to meeting my original objectives (i.e., measuring the total number of cells and total number of TH labeled cells). I also appreciate your comments on the different thresholding methods and threshold correction factors. Manipulating those factors has allowed me to obtain more accurate nuclei segmenation. Thank you very much for your help.:grinning:

1 Like

You can quantify colocalization on stacks on iPad using CoLocalizer app: https://geo.itunes.apple.com/app/colocalizer-for-ipad/id1116017542?mt=8 The app works together with Mac version: https://colocalizer.com/mac/ via iCloud.