QuPath Histology Slide Object detection

Hey QuPath Community,

I have recently started using QuPath for analyzing tendon tissue histology slide images and I am attempting to quantify the following objects and features:

cell number
collagen alignment - organized vs unorganized
Rounding of Nuclei
amount of collagen
vascularity (if present)

I have watched several very informative videos from Pete Bankhead’s YouTube playlist and have been able to implement the onboard cell detection method as well as the onboard ImageJ analysis tools but would like to be more proficient with these tools.

I am working with large svs image files that are stained with Masson’s Trichrome:

The organized collagen would be the concentrated blue pieces up top and the unorganized would be the lighter blue that is slightly below that has a larger amount of cells in it.

Any and all help/tips is greatly appreciated!

It might help to narrow down your questions to what is or is not working, what you have tried, etc. Otherwise the question is a bit broad to approach without a data set to explore.

Most of the available options are described pretty well in the main documents: Welcome to QuPath! — QuPath 0.2.3 documentation

It sounds like some of what you want might require the pixel classifier - but even that might not work for subtle changes in texture. See the post just after yours.

QuPath can also be used to generate training data for deep learning models, the results of which can be imported back into QuPath as annotations or other objects.


Hi @Research_Associate this sequence of post is not very stable :wink:

1 Like

Only so much time before work Tissue separation using QuPath - #2 by Research_Associate