Using CellProfiler 2 plugin in CellProfiler 3 (CellStar)

**tswayne** · March 14, 2019, 3:28pm

To help identify yeast from transmitted-light images, I’d like to try the CellStar algorithm.

However, after downloading the plugins from here to my CellProfiler plugins directory, I still don’t see the new modules. I think the plugin directory is specified correctly because I do see the module from a different plugin that I have in the same directory.

Is this because the original plugin was written for CP 2, and if so, are there guidelines for updating plugins to be compatible with CP 3?

@fafafft, do you have any advice for using CellStar in CellProfiler 3?

Thanks!

**fafafft** · March 15, 2019, 1:11pm

Hi @tswayne!

Great to hear that you are interesting in giving CellStar a try!

The plugin has been written for CellProfiler 2 so I’m not sure that it can work for CP 3 - however maybe it is the same as to be CP 2.3 - time passes so fast : )

Anyway I will take a look at it this weekend and let you know how much work is necessary to make it work.

**tswayne** · March 15, 2019, 2:09pm

Thanks so much, @fafafft!

**fafafft** · March 17, 2019, 11:13pm

Ok, so it works with CellProfiler 3.0.0 and after a small change it also works with CP 3.1.8.

Please use this version: https://github.com/Fafa87/CellProfiler-plugins/tree/sstoma-yit-seg. Use cellstar folder and identifyyeastcells.py from repo instead of the ones from the zip that you downloaded:

Let me know if have any problems, I will be glad to help.

**tswayne** · March 18, 2019, 3:30pm

Thank you! I will try it out soon!

**fafafft** · April 6, 2019, 10:08pm

Did it work? Did it fail?
I hope it did not harm your computer

**tswayne** · April 8, 2019, 4:59pm

Hi @fafafft, I’m sorry, some urgent projects intervened and I didn’t test it yet. Many thanks for your help – I will let you know how it goes!

**noemi** · May 15, 2019, 12:51am

Hi @fafafft,
I just tried CellStar with CellProfiler 3. I do see some errors/warnings when loading pipelines, but they seem to work well with your example data! First pipeline I see that produces good segmentations on bright-field images.
When working with my own images though I still need to figure out the right parameter settings. Was wondering if you have any advice for me on that end. Below is the output using pipeline “StrongEdgesAndWhite_1_Default.cpproj”:

Stack0003_outlinesBoth.tiff (4.1 MB)

Any advice you may have on whether I’m missing any pre-processing steps, which pipeline to use and how to adjust parameters would be much appreciated.

**fafafft** · May 15, 2019, 10:45pm

Hi noemi!

Could you upload the original image and point which object you are interested in?

**noemi** · May 15, 2019, 11:16pm

Sure – below outlines of the cells I would like to identify:

Note that some of the outlines contain multiple cells, that we will want to track individually.

**noemi** · May 15, 2019, 11:19pm

The original image looks completely black:
TimePoint_01_final.TIF (2.8 MB)

I pre-processed it with Fiji to obtain the image from above as input to CellStar.

**fafafft** · May 18, 2019, 11:00pm

Hi!
I have checked your data and unfortunately it will not be possible to use CellStar it in your case. CellStar has two main limitation: cells are expected to be quite regular and roundish and either interior or borders has to be distinctive.

Here are the samples for which CellStar is working well:

As to your data you could try:

reimage the data with different focus plane setting so that cells are more visible (https://blog.cellprofiler.org/2017/06/15/quantifying-microscopy-images-top-10-tips-for-image-acquisition/)
use something like Ilastik to find pixels belonging to cells and then use CellProfiler for the rest of processing (https://blog.cellprofiler.org/2017/01/19/cellprofiler-ilastik-superpowered-segmentation)

**noemi** · May 18, 2019, 11:33pm

OK, I will try that. Thank you!