Trouble Identifying Cells for Measurement

I am very new to using CellProfiler and have just started playing around with it, but have not found a way to have the program properly identify the muscle cells shown from the image.

The closest I’ve come is producing cell-looking shapes as secondary objects based on propagation, with the primary objects as the nuclei (peripherally located at cell borders).
However, the objects identified do not seem to correspond to actual cell shapes at all, and I have not been able to improve on automatic threshold settings.
I’ve attached a screenshot of the images produced from object identification.

I’m continuing to try to educate myself from the tutorials, etc, but would very much appreciate any advice.


Hi,

Glad you are trying out CellProfiler! My first piece of advice is, if at all possible, use a fluorescent marker. Please read this FAQ for more info, since this image looks like it may be tough to extract reliable quantitative measurements from.

But barring that, you should probably begin by inverting your image using ImageMath (CP assumes foreground to be brighter than background), and perhaps you already are. Using the nuclei as primary objects is often recommended, however here, since they are peripheral (and possibly multi-nucleated? hard to tell), it may be better to skip this and try and identify the complete cells as primary objects. It also depends on what you want to measure! What do you want to measure?

For us to help further beyond tips, you should post an original image or two to analyze and your latest pipeline or project file.

Thanks,
David