Analyzing tissue by measuring areas



Hi everyone,

I’ve posted here a long while ago
(Need help with fine-tuning a pipeline (multiple questions: addressing non-uniform stain, deselecting specific areas, morphing a non-completely stained area into one))

(and my pipeline did not resolve the nuclei successfully), so I am changing my strategy, and am looking to measure the number of green only cells, and the number of both green and red cells with the following attached pipeline:

  1. Measure the area of green+ cells: set threshold to green channel, measure the area
  2. Measure the area of red+ cells: set threshold to red channel, measure the area
  3. Mask the green thresholded channel with red thresholded channel (some green cells are not red) -> measure the resulting overlapping area, export the results
  4. Not included in the pipeline: measure the diameter of a few green/or red cells, calculate
    a rough average area of a cell, divide the measured green, or red/green areas by the average cell area to get the number of cells that are both red/green and green only. (Of course Green only = green total - both red/green).

I tried running this pipeline, with the average cell diameter=9.5 (via measurement tool), and the numbers I got are: green only cells = 156, green/red cells=703. However, for my manual count (not 100% precise, but still for now probably more precise than with the pipeline) I got green cells = 50, green/red cells = 867.

So, I was also wondering if there’s a better way to calculate the average cell area? Also, is it possible to select a region of interest in Cell Profiler? (I cut out the ROI in photoshop since in the original image there are red + green areas in the basement membrane that I’m not analyzing). I’m attaching the cropped and one original image.

(Originally, I’ve tried using a strategy of IDing nuclei as primary objects, and overlaying them with the thresholded red layer and green layer. However, the nuclei are very densely packed so their resolution proved a bit tricky so far).

Any other tips greatly appreciated!

05032018_Measure_By_Area_v2.cpproj (657.5 KB)



You could add an ROI with the IdentifyObjectsManually; alternately, you could try to Threshold the DAPI channel, then use MaskImage on the green/red images and doing the rest of your workflow from there.

I’m most troubled by the fact that the relative ratio of your green vs green/red cells is so different (rather than the actual counts)- if this persists after adding some kind of masking strategy I’d either

  1. Measure the diameter of green and green/red cells to see if they tend to be different sizes
  2. Make sure that your thresholding is actually being applied how and where you want- are there dim cells being missed, etc?

Good luck!