CellProfiler- building pipelines

Hi, I am an undergraduate student in Univ of Sheffield, in UK.

I am doing a dissertation on evaluating cell tracking tools, and CellProfiler is one of my test candidates.

I have been trying to use the tool to track cells in images(jpeg). I tried creating the pipeline myself but found some difficulties.

I need to identify two different types of cells; ones in green and ones in red.

I have attached my .cp file and one of the image file.

Could I please get some help in building appropriate pipeline to track the cells in a given area(using crop).

Thank you.

p.s. I have been trying to understand the tracking methods within the tools, however I just did not really understand them. I am not doing a biological science or mathematics but im studying enterprise computing; I have a bit of programming language understanding but not at the lowest level.
It would be very very very helpful if I could get some basic description of methods and what they are advantageous at. Thanks.



DefaultOUT.mat (47.9 KB)

Hi,

I’ll address your issues separately:
Identification

  • We recommend acquiring your images as TIFs or PNGs since they do not suffer from compression artifacts which degrade image quality as JPEGs do.

  • Your cells have very heterogeneous borders which will probably made identification somewhat problematic. Do you have a nuclear stain which you can use? If so, you can use IdentifyPrimaryObjects to detect the nuclei and then IdentifySecondaryObjects to find the cell body, both the red and the green.

Tracking

  • Are your cells moving quickly or slowly? If there is a good amount of overlap between cells from frame to frame, you may find the Overlap method to be straightforward.

  • Whatever tacking method you chose, whether one of CellProfiler’s methods or a completely different program altogether, tracking results are only as good as the cell detection. I strongly suggest that you optimize the cell detection as mentioned above before you proceed to tracking issues.

Hope this helps!
-Mark