Problems with different nuclei sizes and intensities

Hi,
I just started using CellProfiler the other day and still have some problems figuring out the right settings. I did a basic Hoechst staining and wanna count the number of nuclei as well as the average nuclei size. I have two main problems:

  1. the nuclei size on my pictures are ranging from 15µm to 80µm and I haven’t figured out a way yet I can count them both, without either getting multiple counts for the big nuclei or only 1 count for multiple small nuclei. I used several setting including discarding objects outside of range, changing nuclei diameter etc…

  2. I have a big heterogeneity in my staining intensities meaning some of the nuclei are really bright others are rather dark. The dark ones are not getting counted.
    Is there any way I can set the cell profiler to include all of the nuclei.

Thanks for your help.

I attached a sample picture

Hi Angela,

One item that may help is the fact that automatic values of the smoothing filter size and maxima suppression sizes are linked to your input for the lower diameter bound. So you can try setting the permissible diameters (you can use “Tools > Measure length” from the display window to measure with) and then uncheck the “Automatically calculate size of smoothing filter?” and “Automatically calculate minimum allowed distance between local maxima?” settings and start adjust them independently to find what settings work best for your images.

Also, using the Laplacian of Gaussian method of distinguishing clumped objects sometimes works better to delineate blobs of widely varied sizes, if used in conjunction with the suggestion below. However, it can be tricky to optimize the settings in this case.

[quote=“Angela”]2) I have a big heterogeneity in my staining intensities meaning some of the nuclei are really bright others are rather dark. The dark ones are not getting counted.
Is there any way I can set the cell profiler to include all of the nuclei.[/quote]

I would suggest using Otsu global as the thresholding method, with 3-class thresholding and the middle class set to foreground. This should capture more of the dim nuclei.

Hope this helps!
-Mark