Extract morphological features with Python library

Sample image and/or code

Untitled

Analysis goals

How to extract morphological features (such as Granularity_9_MaskedEosin, Texture_Contrast_NucleiNeighborCount_100_90) from pathological images (like in the image) with python library?
I need to use this CellProfiler · PyPI

If you’re interested in trying to do this analysis with the graphical user interface of CellProfiler, I recommend watching this CellProfiler introductory workshop, which is available on the Center for Open Bioimage Analysis YouTube Channel. You can download the corresponding written tutorial on Translocation from the CellProfiler Github page. If you have questions while building a pipeline to analyze your images, please do reach back out on the forum!

Best,
Pearl

1 Like

Thank you, Pearl!
Is there a way to use python api directly? Like how does sklearn library?

In the Imaging Platform, we use the GUI version to build our pipelines and then Distributed CellProfiler to run CellProfiler on Amazon Web Services when we need to analyze large datasets. We use the pip installer sometimes for installing and then running CellProfiler from source (Source installation (OS X and macOS) · CellProfiler/CellProfiler Wiki · GitHub), but otherwise we don’t use the Python api internally.

I would be curious to hear from others in the community if they’ve used CellProfiler via Python api. If you’re looking to analyze images via Jupyter notebook, for example, I think we’d generally recommend that you use the various numpy, scikit-learn, scipy packages directly.

You CAN use CellProfiler as a Python package; I’ve linked to our documentation on it below. It’s a good way to get up and running quickly with an existing CellProfiler workflow.

However, my advice would be that there aren’t that many situations where I would recommend that as your best option- in general, if you need a string of modules/a whole pipeline, I’d just recommend you use our headless run tools, and if you need only one module, I’d just grab the underlying skimage and/or centrosome code and save yourself the “overhead” of all CellProfiler’s requirements. Nevertheless, if you do choose to do it, here’s how.