How to detect cells in IHC slides if the nucleus is obscured by DAB?

[Vetdocjim1]

Dear colleagues:

I am working with on b-catenin IHC project. In some cases, the positive signal obscures the nuclei:

Ultimately my goal is to assign an H-score to the cells, but I have not been able to determine how to configure QuPath to identify the cells if the nuclei are obscured by DAB. I guess one approach would be to do 2-color IF and use DAPI in one channel.

I would appreciate your thoughts.

Thanks!!

Jim

[micros...@gmail.com]

2 color IF would help, though your results may be interesting depending on the shape of the cells.  As long as you are okay with a certain amount of error it should be find (cell expansion is not usually defined by the surrounding area, though you could try the Cell+membrane detection.  I have not had much luck).  Another suggestion would be to increase the staining time of your hematoxylin in order to make it darker, which would allow you to more easily use the Hematoxylin OD to define your cells.  If reasonably available, IF is probably the way to go though.

Another option if you can avoid cell counts is area based measurements like SLICs, or Region Identification/Create cytokeratin annotation.  The latter gives you separate annotations that can allow you to use different thresholds for your cell detection in different areas, which may let you count the cells in each type of area (dark DAB vs light DAB) more accurately.

[Pete]

I'd certainly try Analyze -> Cell analysis -> Cell + Membrane detection.

It only works with hematoxylin & DAB staining, but can create cell objects based purely on membrane staining even if no nucleus is visible.  It was created when working with HER-2 staining.

For image analysis in general, I'd much rather work with fluorescence images, and would ideally like the 2-color IF idea.  Then the general Cell detection command can be used - although as microscopyra says, the cells are created based only on nucleus detection and distance, and without actually using the information in the other channel.  This might still be ok though, and give more options for a day in the future when QuPath may have a new and improved cell detection command working for fluorescence images...

Finally, when using your current image with Cell detection you should see an option in the dialog box Exclude DAB.  This tells QuPath to avoid detecting a nucleus if it finds a stronger signal in the DAB.  It could be a bit hit-and-miss (limited by how well the stains could be separated), but is probably worth a try.